ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
1
Chapter V Bipolar Junction Transistor Circuits
The bipolar junction transistor (BJT) is very versatile three-terminal device that has been extensively used within integrated circuits since its invention in the late 1940s. Over the years, these multipurpose BJT devices have served as amplifiers, switches, and temperature sensors in many different types of analog and digital chips. Nowadays they are still popular, especially in amplifiers requiring a wide frequency range and in voltage/current reference circuits. Since BJTs are available as inexpensive individually packaged components, you can also encounter them in circuits assembled on printed circuit boards for machine control applications and instrumentation equipment. In this chapter, you will learn the BJT’s operation and small-signal model, as well as how to use BJTs to design circuits. V.1. BJT Fundamentals. In this section, we will discuss the main principles behind the operation of the BJT. The NPN bipolar transistor consists of two P-N junctions, as depicted in Figure 5.1a. Essentially, the bipolar device is composed of two back-to-back diodes with different doping profiles, and its three terminals are conventionally labeled as E (emitter), B (base) and C (collector). The emitter is the most heavily doped terminal and it emits the majority of the carriers that flow through the base-emitter junction. The reason why the emitter is the most doped region is that high doping levels promote current flow. When the BJT is operating as an amplifier, the base-emitter diode is usually forward biased such that a large amount of current can be flow through this junction. Since the base is a very thin layer, most of the carriers from the emitter that are injected into the base have enough energy to travel through the base and are accumulated in the collector. This mechanism requires the collector terminal to have sufficiently high potential energy. For that purpose, the voltage at the collector terminal must be more positive than the voltage at the emitter, where the required difference is typically more than 0.3 V. A small number of carriers are recombined in the base, which generates a small base current, but most of the carriers reach the collector terminal where they give rise to output current. The number of electrons reaching the collector terminal is determined by the base-emitter potential in a similar manner as the current flowing through a forward biased diode. Thus, the output current (iC) is controlled by the base-emitter voltage, leading to an operation resembling a voltage controlled current source (or transconductance amplifier) since the input is voltage and output is current. Nevertheless, it is often modeled as a current controlled current source based on the direct dependence of the collector current on the base current.
iC
iE
iB- - - - -- - - - -- - - - -- - - - -- - - - -
- -
- - -
- -
+ + +
+ + +
+ + +
Electrons
Holes
+VBE
-
+VCE
-
E
B
C
iB
iCiE
(a) (b)
Fig. 5.1. a) NPN transistor and b) its symbol. The applied base-emitter voltage determines the current flowing across the emitter-base junction, where the current flow is mainly attributable to the electrons originating in the emitter because the emitter doping concentration (> 1017 electrons/cm3) is much higher (typically around 100 times) than that of the base. To focus more on the characteristics and functionality of the BJT with regards to designing circuits, the physical principles behind its operation are not discussed here. If you are interested in this topic, you can find excellent explanations in several electronics and semiconductor device physics books. The emitter current follows the same v-i relationship as that of the typical diode:
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
2
1iE
th
BE
V
v
SE eI , (5.1)
where ISE is the saturation current for the emitter terminal, is a fitting factor (usually between 1 and 2), and Vth is the thermal voltage proportional to the temperature. The saturation current is dependent on the emitter area as well as other physical parameters, and its value is strongly affected by temperature variations. Typical ISE values are in the order of 10-12 − 10-15 A. ISE approximately increases by factor of two when the temperature increases by 10 degrees. The thermal voltage is temperature dependent as well, and it can be calculated by using the following expression:
q
kTVth , (5.2)
where k is the Boltzmann constant (=1.38×10-23 J/K); T is the temperature in degrees Kelvin (300° K = 27° Celsius), and q is the fundamental charge of the electron (=1.6×10-19 Coulomb). At room temperature (300° K), the thermal voltage is roughly 26 mV. The symbol for the NPN transistor is shown in Figure 5.1b, in which the arrows indicate the directions of the currents at the terminals. As you can observe in equation 5.3, the collector current exhibits a similar behavior as the emitter current.
1eIi th
BE
V
v
SC (5.3)
The only difference between iC and iE is the slightly different value of the saturation current IS. Since the emitter and collector current are very similar, we use a factor for denoting their relationship as follows:
1I
I
E
C . (5.4)
An approximation of ≈ 1 can often be made in first hand calculations. The thinner the base is, the smaller are the carrier recombination and the base current, and therefore approaches one (IC = IE) as the base current approaches zero. It should also be noticed that the transistor is not generating any energy by itself, hence the current flowing through the emitter must be equal to the addition of the base and collector currents as stated in the following equation:
BCE iii . (5.5)
According to equations 5.4 and 5.5, we can find the relationship between base current and collector current:
11
B
C
I
I . (5.6)
This expression is known as the common-emitter current gain factor. represents the current gain of the amplifier for the case in which the input signal is applied to the base terminal and the output current is flowing at the collector terminal of the transistor while the emitter terminal is grounded,.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
3
V.2. DC Characteristics of the BJT A bipolar transistor’s IC-VBE curve is obtained by sweeping the base-emitter voltage. This plot depicted in Figure 5.2 is known as the DC input characteristics. For negative voltages, the emitter-base diode is reverse biased and it operates as a very large resistor, such that the emitter current is around -IS (less than -0.1 pA). If the diode is forward biased, then the collector current is determined by expression 5.3.
IC (A)
VBE (V)0.6 0.8
QCQ
VBEQ
0.40.20.0-0.2-0.4
Fig. 5.2. DC input characteristics of a BJT.
For base-emitter voltages below 0.5 V the collector current is very small, which is typically less than 1 A. It will be evident in the next sections that the BJT is often operated with relatively large DC collector current and base-emitter voltages VBE > 0.5 V to ensure that the DC current is larger than the AC current to be processed. Once a base-emitter voltage is selected, the collector current directly depends on this selection, and the operating point Q (also called the quiescent point) is therefore identified by the VBEQ and ICQ combination. The collector-emitter voltage VCE can be swept for a given VBEQ, leading to the transistor’s DC output characteristics shown in Figure 5.3. With small VCE (< 300 mV), the potential at the collector terminal is not high enough to attract the carriers from the emitter, resulting in small collector current. This region of operation is known as saturation region. If the transistor is operating in saturation region, the expression for the collector current given in equation 5.3 is not valid. For higher collector-emitter voltages (> 0.5 V) most of the minority carriers (electrons) flowing through the emitter-base junction are attracted at the collector terminal. In this case, the BJT is said to be in the linear region, which is also called active region. If VBE changes while the BJT is in the linear (active) region, the collector current will follow the exponential dependence on VBE given by equation 5.3. Finally, there is another region of operation referred to as cutoff region, which is the state that occurs when the emitter-base junction is reversed biased (VBE < 0.7 V). When the BJT is in the cutoff region, then the base, collector, and emitter currents are almost zero. When identifying the regions of operations and making the associated assumptions, keep in mind that the aforementioned voltages vary depending on the device materials, fabrication characteristics, and manufacturer. For example, you could encounter cutoff voltage values in the 0.5 V to 0.8V range, or even outside of this range. Furthermore, if devices are biased under conditions at the boundaries of the regions, their properties are usually not well defined. For instance, you cannot expect the IC-VBE characteristics to be perfectly exponential when the BJT is operating at the edge of the linear region with a collector-emitter voltage only slightly above 0.5 V.
C
VBE1
VCE
VBE4
VBE2
VBE3
VBE0
Fig. 5.3. A BJT’s DC output characteristics for a family of VBE voltages, where VBE0 < VBE1 < ... < V BE4.
Notice in equation 5.3 that the fundamental equation for the transistor’s behavior is non-linear. When AC signals are present in transistor circuits and the goal is to calculate the resulting output signals, then the equations derived from equation 5.3 usually do not have a closed form solution. Even if the solution can be found, the equations will be quite complex, making it is very difficult to get sufficient insights into a circuit’s behavior. In most of the cases, computer based simulation methods are used for plotting the output waveforms as function of the input waveforms.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
4
V.3. Small-Signal Model for the BJT: A Linear Approximation. Since the complexity of the non-linear systems increases with the number of transistors in the circuits, we have to limit our analysis in order to derive simple expressions based on the assumption that the AC input signals are small. The idea is to make reasonable approximations that give us more insights into the operation of the circuits. Although the accuracy of the results will be limited, the solutions are reasonably close to our targets with errors within ±10 % in many cases. Once the circuit is designed, we have to simulate it and re-adjust some of the parameters to obtain more accurate results. The design of circuits is usually an iterative process due to the limited accuracy of the theoretical equations and sometimes even due to the discrepancies of the simulation models. A typical amplifier combines DC and AC signals at the input of the transistor. The circuits used for this operation will be discussed later on, but for the following analysis we assume that the base-emitter voltage has two components as shown in Figure 5.4.
RC
vbe
+-
vCE
+
-
iC
Fig. 5.4. Typical common-emitter amplifier. The base-emitter voltage has both DC (operating point) and AC (signal)
components.
The voltage applied to the base-emitter terminal (VBE + vbe) is converted into a current iC by the transistor. Notice that the transistor is a base-emitter voltage to collector current converter. Since the relationship is exponential, any small signal applied to the base-emitter terminals is mapped into large collector current variations, as can be seen in Figure 5.4. This is the main principle behind the BJT operation. It is worth to make a couple of important observations at this stage: i) The larger the selected bias (time-invariant) current ICQ is, the larger is the AC current generated by the
input signal. Or in other words, larger DC bias current ICQ will result in higher gain during the conversion of an AC input voltage to an AC output current.
ii) If the AC (time-variant signal) input signal vbe to be processed is small, then the exponential collector current can be predicted by a linear approximation, which will be apparent shortly. Figure 5.5 visualizes the input-output relationship for a BJT with a set DC bias and an additional sinusoidal AC input signal with small amplitude at the base terminal. You can also recognize in the plot that the superposition principle holds in such a situation since the output current is simply the addition of the AC and DC input voltages. However, in many practical applications we are interested in processing AC signals, and only select the DC bias conditions to guarantee the appropriate transistor properties as discussed later in this chapter.
Tim
e
Fig. 5.5. Relation between iC and VBE when a small AC input signal is applied to a BJT device that is biased at a DC
Q-point in the linear region.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
5
When a combination of DC and AC signals is applied to the transistor’s base-emitter junction, we can write the expression for the collector current based on equation 5.3 as
th
be
th
be
th
BE
th
beBE
V
v
CQV
v
V
V
SV
vV
SC eIeeIeIi
1 , (5.7)
where ICQ is the DC component (collector bias current) of the output current. In this expression we assumed that the transistor is operating in the linear region and that VBE > 4·Vth ≈ 100 mV. As will be elaborated in the following sections, this condition is normally satisfied because the DC base-emitter voltage is usually greater than 500 mV in linear amplifier applications. In the previous equation, the small signal we want to amplify is denoted as vbe in lower-case letters. Since the equation is a smooth function of vbe, it can be expanded in a Taylor series around the operating point Q (VBE, ICQ), which is:
....62
32
th
beCQ
th
beCQ
th
beCQCQC V
vI
V
vI
V
vIIi (5.8)
The first term in this equation corresponds to the DC current component. The second term is the desired collector current component, which is linearly related to the incoming AC signal vbe. The remaining terms lead to the undesired high-order harmonic distortion components due to the transistor’s non-linear characteristics. To get more insights into the effects of the higher-order terms, let us consider the case of a sinusoidal input signal vbe = Vpk·sinot. Using trigonometric properties and considering the first four terms of equation 5.8, it can be shown that the collector current can be also written as
tVV
V
V
I
tVV
V
V
I
tVV
VI
V
I
V
VIIi
opkth
pk
th
CQ
opkth
pk
th
CQ
opkth
pkCQ
th
CQ
th
pkCQCQC
3sin24
2sin4
sin8
4
2
2
2
.
(5.9)
The second term in this equation shows that the transistor non-linearity creates a small change of the DC current. Furthermore, it reveals the generation of undesirable signals called harmonic distortion components, which are located at frequencies that are multiples of the fundamental signal frequency ωo. The ratios of the amplitudes of these unwanted components to the desired signal are defined as the harmonic distortions. These ratios are frequently used to measure the quality of output signals and to provide quantitative information to the engineers who design systems based on the output signals of analog amplifiers. The first two harmonic distortion quantities are defined as
th
pk
pkth
pk
th
CQ
th
CQ
pkth
pk
th
CQ
o
o
V
V
VV
V
V
I
V
I
VV
V
V
I
Amplitude
AmplitudeHD
4
1
8
4
@
2@2
2
, (5.10a)
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
6
2
2
2
24
1
8
24
@
3@3
th
pk
pkth
pk
th
CQ
th
CQ
pkth
pk
th
CQ
o
o
V
V
VV
V
V
I
V
I
VV
V
V
I
Amplitude
AmplitudeHD
. (5.10b)
In the above equations, the approximations can be made because the second term within the parentheses of the denominator is much smaller than the first term under the assumption that the AC signal current is small compared to the DC current. A piece of information that will help you to interpret the equations for the harmonic distortion quantities is that the unwanted harmonic distortions reduce the quality of the output signal. Thus, it is desirable to maintain a small input signal amplitude (Vpk) to reduce distortion components and to obtain a linear output signal. Although the largest harmonic distortion is the second-order component (HD2), it can often be suppressed by properly designing the circuit as a fully-differential system with two symmetrical signal paths that process the same signal with opposite polarity. The outputs of a fully-differential circuit can be subtracted, leading to significant cancellation of even-order distortion components (HD2, HD4, etc.) without affecting the fundamental signal. Cancellation methods for HD3 and other odd-order distortion components is a lot more complicated, which is why it is less frequently done to optimize linearity performance. Notice that if we want to reduce the third harmonic component down to 1% of the fundamental component (HD3 = 0.01), then the peak value of the incoming signal Vpk must be limited to Vth/2 = 13 mV. For this reason, we will often limit the signals directly at the base-emitter terminal to vbe < 10 mV such that the circuit can be regarded as a “linear” system. Using this approximation, the resulting model for the transistors AC behavior is termed small-signal model.
iC
0 0 0
HD2HD3
0
iC
0time
iC
0 0 00
iC
0time
HD2 HD3
(a) (b)
Fig. 5.6. Frequency spectrum (top traces) and corresponding transient waveform of a transistor’s output current (bottom traces) with harmonic distortion components: a) with severe non-linearity, b) with better linearity.
Figure 5.6a shows the distorted output current of a BJT in the frequency and time domains. Since the distortion components can be very small, it is usually better to observe the output current using a logarithmic scale. In general, the amounts of distortion are conventionally reported as a percentage or in decibels. For example, the HD3 of 0.01 (1%) corresponds to -40dB since HD3dB = 20·log(HD3). If the magnitudes of the harmonic components are plotted
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
7
in dB, then HD2dB, HD3dB,… can be easily determined as the difference between the fundamental component and the components at 20, 30,...; respectively. In comparison, Figure 5.6b depicts the output current spectrum and corresponding transient waveform with reduced distortion components. The improvement can be achieved by lowering the input signal amplitude. Sometimes, selection of non-ideal DC bias conditions creates distortion, especially when the VCE > 0.3V condition is not guaranteed for the maximum output signal swing, or when the output signal swing is so large that it reaches the supply voltage VCC which leads to clipping of the signal. While designing analog circuits, you should always check your transient output signal for the maximum expected input signal amplitude to verify that it does not become distorted due to poor choices of component values or DC bias voltages and currents. Very simple but useful equations can be derived by assuming that the magnitude of the AC signal is smaller than 0.4·Vth (≈ 10 mV). For this case, the collector current according to equation 5.8 can be approximated as a linear function of the AC base-emitter voltage given by
beth
CQCQC v
V
IIi
. (5.11)
Fundamentally, the collector current has two components: the bias current ICQ and the AC current component that is linearly related to the base-emitter signal voltage. Usually, the AC component is the information that we want to be processed. The parameter that is mapping the AC input voltage vbe into the AC output current is the first derivative of the iC-VBE curve evaluated at the operating point (ICQ, VBEQ). This parameter is defined as the small-signal transconductance gain:
th
CQ
QBE
Cm V
I
v
ig
. (5.12)
With units of A/V (Siemens, S), the small-signal transconductance of the transistor represents the slope of the linear approximation at the operating point as annotated in Figure 5.7. From this plot, you can also intuitively verify the mathematically expected increase of harmonic distortion components for larger input signal swings. The tangent line that represents the linearization agrees well with the exponential iC-VBE relationship around the Q point, but the actual output current iC deviates from the line more as the vBE amplitude is increased around VBEQ. This deviation causes distortion of sinusoidal signals due to the non-linear input-output characteristics.
.
QBE
Cm v
ig
Fig. 5.7. Linear approximation for the transistor’s input characteristics at the operating point Q. Around Q, the iC-
VBE non-linear relationship is approximated by a straight line with a slope given by ICQ/Vth. A simple transistor model can be obtained based on the above assumptions and analysis. In the actual BJT, there is a forward biased diode connected between base and emitter. The base-emitter voltage controls the collector current that can be modeled as a voltage controlled current source, resulting in the small-signal model of the BJT that is depicted in Figure 5.8a. As mentioned earlier, the output current has both DC and AC components. Hence, the diode at the base input should be modeled as a DC voltage source in series with the diode’s resistance r, where the base-emitter resistance ris 1/g. The transconductance g is determined by differentiating the base current with respect to the base-emitter voltage around the operating point as follows:
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
8
th
BQ
QBE
B
V
I
v
i
r
1g
. (5.13)
E
IC+gmvbeiB
B
C
ie
ic
r
+0.7V
-
(a) (b)
Fig. 5.8. Small-signal model for the BJT with forward biased base-emitter junction: a) the base-emitter diode is shown, b) the base-emitter diode is represented by a DC voltage source and a series diode resistance.
Notice that both DC and AC signals are applied to the BJT’s input at the same time, and that the small-signal parameters r and gm depend on the DC operating point since these parameters are determined by the collector and base current derivatives evaluated at Q. For small signal conditions (|vbe| < 10 mV), the transistor operates as a quasi-linear device. Under these conditions, we can use the superposition principle that applies to all linear systems, and analyze the circuit for two different cases: DC signals only, AC signals only. You will see in the following discussions that splitting the analysis into two parts simplifies it. But at the same time, keep in mind that the output voltage consists of the two components and that many AC parameters such as gm and r depend on the DC operating conditions. V.4. Transistor Model and Design Considerations. V.4.1. DC analysis: operating point and definition of the small-signal parameters. The first analysis is carried out only for DC conditions. Hence the AC signal sources are set to zero by replacing AC voltage sources with short circuits and AC current sources with open circuits in the diagram. Next, the voltages and currents that define the operating point Q can be calculated in order to find the small-signal parameters (e.g. r and gm). Alternatively, the appropriate operating point and bias conditions can be selected to meet given small-signal parameter requirements. The DC analysis of the circuit shown in Figure 5.4 is carried out by making vbe = 0, which leads to the diagram in Figure 5.9 and following analytical results:
th
BEQ
V
V
SCQ eII . (5.14)
From this equation, the VBEQ voltage controls the collector current and therefore the operating point on the input characteristic curve. Usually VBE is in the 0.5-0.8 V range.
VCC
RC
+-
VCEQ
+
-
ICQ
Fig. 5.9. Diagram for the DC analysis of the common-emitter amplifier.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
9
The collector-emitter voltage VCE depends on both collector current and RC. From the circuit shown in Figure 5.9 we can find that
CCQCEQCC RIVV . (5.15)
There is a linear relationship between VCE and IC, termed load line. For a given VBEQ, both equations 5.14 and 5.15 define the operating point Q (VBEQ, ICQ, VCEQ) of the amplifier as shown by the plots in the Figure 5.10a.
(a) (b) Fig. 5.10. BJT output characteristics and load line: a) the operating point Q is defined by the selected combination of
VBEQ, ICQ, and VCEQ; b) effect of different load resistors RC on the Q point when VBE is fixed. Since both equations 5.14 and 5.15 affect the operation of the amplifier, the operating point Q is determined by the intersection of the transistor output characteristic associated with VBE and the load line. The load line depends on the power supply VCC and RC. In fact, its crossing point on the x-axis is equal to VCC. Furthermore, the slope of the linear equation 5.15 in the iC-VCE plane is given by -1/RC. Therefore, VCEQ decreases as the resistor value is increased, which can be seen in Figure 5.10b. Notice that increasing RC reduces the slope of the load line, pushing the operating point Q closer to the non-linear regime of the transistor on the iC-VCE plane when VCE < 0.3 V. This is an unfavorable design condition because the output could be very non-linear. It is always good practice to locate the operation point Q within the flat region of the red curve in Figure 5.10b such that VCE > 0.3V under all conditions, even when the output voltage at the collector reaches its minimum point due the processed AC signal that will be superimposed with the DC voltage VCEQ. When an AC signal is added to the DC base-emitter voltage, the overall input voltage is modulated by that signal as shown in Figure 5.10. Accordingly, the operating point moves on the load line to follow the signal. By this process, the signal variations at the base-emitter junction are mapped to the VCE-axis, generating the collector-emitter output voltage. Since slope of the load line is -1/RC, a larger resistor value at the collector terminal will lead to a more amplified output signal. However, RC cannot be increased unconditionally because the operating point Q moves towards the non-linear saturation region where large harmonic distortion components might be generated.
iC
VCE
Q
VCEQ
ICQ VBEQ
VCC
VCC/RC
0.3V
linear range
vbe-peak
vce-peak
Fig. 5.11. Modulation of the operating point (Q) in response to an applied AC base-emitter voltage, where the Q
point moves along the load line as indicated by the blue arrow.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
10
Selection of a proper operating point is one of the most critical considerations during the design of linear amplifiers. The operating point Q must be selected based on the following criteria: i) The Q point must be able to accommodate the signal variations. Since both DC and AC signals are present
at the same time, the overall collector-emitter voltage must vary such that 300 mV < VCEQ - vce-peak < VCC. Otherwise, the transistor will enter the saturation region or the signal will be limited (clipped) by the supply voltage VCC.
ii) The AC parameters r and gm are functions of IBQ (= ICQ/) and ICQ, respectively. The larger the DC current gain of the transistor () is, the smaller will be the base current, and the larger will be the base-emitter resistance r. Often, the collector current ICQ is directly computed from the required small-signal transconductance gm that determines the amount of AC collector current generated by the AC input signal.
V.4.2. AC analysis: the -hybrid model for BJTs. The second analysis is carried out only for the AC signals, while the DC voltage sources are shorted and the DC current sources are considered as open circuits. This is the analysis that defines all AC parameters such as input and output impedance, current and voltage gain, as well as the operating frequency range. A simplified AC small-signal model of the amplifier in Figure 5.4 is displayed in Figure 5.12a.
RC
vo
+
-
ic
B
E
C
E
rgmvbe
ib
B
RC
ie
voC
vbe
(a) (b)
Fig. 5.12. Schematics for AC analysis: a) AC equivalent diagram of the amplifier, b) equivalent circuit with the -hybrid model for the BJT.
Solutions for AC equivalent circuits can be obtained by using fundamental circuit analysis. In this circuit, the AC output voltage is obtained by multiplying the collector current with the load resistance RC. Since the base-emitter voltage in this circuit is equal to the AC input signal, the expression for the voltage gain becomes
th
CCQCm
be
o
V
RIRg
v
v . (5.16a)
The voltage gain depends on the DC voltage drop (ICQRC) across the load resistor RC. As observed during the load line analysis, a larger load resistor RC increases the small-signal voltage gain is, but also notice that what really matters is the product ICQRC. At room temperature, the thermal voltage is around 26 mV. Hence, the previous equation can be rewritten as
CCQth
CCQ
be
o RIV
RI
v
v40 @ room temperature. (5.16b)
This result will be routinely used in the following sections. On the other hand, it should be mentioned that the input impedance at the base terminal is r= Vth/IBQ. The smaller the base current, the larger is the input impedance. In future analyses you will see that the input impedance has a more significant impact on the overall gain when the signal voltage source at the base has finite output impedance. V.4.3. Inclusion of the transistor's collector-emitter resistance. A non-ideal effect present in the BJT operation is the collector current modulation due to the collector-emitter voltage. The collector current increases for large collector-emitter voltages because more carriers are attracted to the collector due to the higher electric fields. If the
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
11
base-emitter voltage is fixed while the collector-emitter voltage is swept, the transistor’s output characteristics in Figure 5.13 is obtained. This figure visualizes this sweep of VCE for four different VBE voltages, where the finite slopes above VCE > 0.3V are caused by the finite collector-emitter resistance rce. Extrapolating the current-voltage characteristics for negative collector-emitter voltages gives us the so-called Early voltage Vearly, which is the point where they intersect the x-axis. The Early voltage has little sensitivity to the bias conditions, and it depends strongly on the specific process technology used to fabricate the BJTs. Typical Early voltages have magnitudes in the range of 50-100 V for standalone BJT devices. The slope of the iC-VCE characteristics is defined as the transistor’s output conductance given by
early
CQ
earlyCEQ
CQ
QCE
C
ceo V
I
VV
I
v
i
rg
)(
1. (5.17)
Fig. 5.13. iC-vCE characteristics of the BJT. The slope of the curves represents the transistor’s output conductance go. The approximation in the previous equation is under the assumption that VCEQ << Vearly, which is a realistic assumption for discrete transistors but not necessarily for BJTs within integrated circuits. An improved model for AC signals is depicted in Figure 5.14a that takes the effect of the current modulation due to VCE into account. Here, the collector-emitter resistor (rce = 1/go = Vearly/ICQ) has been added. It will be evident in the following sections that this model is very useful for the analysis of common-emitter topologies wherein the emitter is connected to a DC voltage source or ground. However, for the analysis of some other topologies it is more convenient to use the T-model presented in Figure 5.14b.
re
ie
ieib
C
rce
ic
(a) (b)
Fig. 5.14. Small-signal models for a BJT: a) -hybrid model, b) T-model. V.4.4. The T-model for BJTs. The T-model in Figure 5.14b is based on the collector current representation in terms of the emitter current rather than in terms of the voltage vbe that is generated from the base current in the -model. If the transistor is now analyzed at the emitter, the current flowing through the emitter-base diode is the entire emitter current. Therefore the diode’s conductance seen from the emitter terminal is given by
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
12
th
EQ
QBE
Ee
V
I
v
ig
, (5.18a)
and the emitter resistance is
1
r
I1
V
I
Vr
BQ
th
EQ
the . (5.18b)
The collector current must be modeled as current-controlled current source in the T-model, which is iC = ·iE. Another condition that must be added to this model is the KCL condition iE = iC + iB. By taking these equations into account, the equivalent T-model shown in Figure 5.14b is fully explained. As in the π-hybrid model, the rce is also included between the collector and emitter terminals. Both models will be extensively used in the following sections. The models represent the same set of equations, and the final numerical results will be independent of the model used in the analysis. However, one of the models often allows easier algebraic analysis for a given circuit. When you investigate a new circuit with one model and experience difficulties to obtain solutions in the first attempt, it is usually helpful to restart the analysis with the other model. In general, the -hybrid model is very popular, but for some topologies it is simpler to gain insights into the circuit’s performances by using the T-model to derive expressions. V.4.5. Practical transistor limitations. The aforementioned models are valid if and only if the transistor is operating in the linear region. Some practical constraints of the transistor model are: i) Transistors operating with very small collector-emitter voltage (VCE < 300 mV) will operate in the non-linear
saturation region. If the base-emitter voltage is large enough (VBE ≈ 0.5 – 0.8 V) under this condition, then the input diode is on and huge base-emitter current can be generated. The collector current however is almost linearly related to VCE since the electric field between the base and emitter terminals is not strong enough to attract the minority carriers that are traveling throughout the base. As a result, the collector current can be very small compared with the emitter current, and both the and current gain factors reduce drastically when the transistor is operated in the saturation region. Since the transistor is quite inefficient if operated in this region, it is recommended to maintain VCE > 300 mV under all possible operating conditions when designing linear amplifiers.
ii) Although the exponential behavior of the collector current is valid over several decades (in many cases from 0.1 A until 10 mA), the collector current is limited due to non-ideal effects such as parasitic resistances embedded in the transistor, velocity saturation of the carriers, and other second-order effects. These effects reduce the amplification efficiency of the BJT at high current levels, limiting its current-gain factor especially in power amplifier applications. Figure 5.15 illustrates the typical evolution of the current-gain β as the collector current increases. In this figure, you can also observe the sensitivity of β to temperature changes.
Fig. 5.15. Typical variations of as a function of the collector current at different temperatures.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
13
iii) Since most AC parameters such as r, gm, and are quite sensitive to temperature variations, it is critical in many practical applications to improve the design’s robustness to these variations as much as possible by minimizing the circuit’s sensitivity to temperature changes. The normalized sensitivity of a function H to variations of a parameter T is defined as
T
TH
H
T
H
H
TS H
T
. (5.19)
Here, function H could be the equation that describes a performance parameter such as the amplifier’s gain, and parameter T could be the temperature or another parameter such as β that changes due to manufacturing variations. When equation 5.19 is too complex to be evaluated analytically, then simulation programs can be used to find the sensitivity by repeating simulations with incremental changes or wide-range sweeps of parameter T, while recording the corresponding values of the output parameter H under investigation. The sensitivity function represents the ratio of the variations of both H and T with respect to the nominal value, i.e. it is normalized. Sensitivity of 1 means that the normalized variation of T affects the normalized function H in the same proportion. For instance, if a 10% temperature change from 25°C to 27.5°C generates a 10% increase of from 150 to 165, then we say that the normalized sensitivity is 1. iv) Very large base-emitter voltages produce very large collector currents, which will cause the temperature of the
BJT device to change as a result of self-heating. Additionally, drastic increases of the collector currents occur when voltages exceed the breakdown voltages in the junctions. For example, the breakdown voltage VCEB labeled in Figure 5.16 depends on the technology with which the BJT is fabricated, and its value can be found together with the specifications of the other breakdown voltages in the device datasheet provided by transistor’s manufacturer. As you can see in the figure, VCEB also has some dependence on VBE, which is why it is good practice to list and pay attention to the operating conditions under which specifications are reported in the datasheet. In conclusion, it is strongly advisable not to reach the specified voltage and power breakdown limits.
VCE
VBE1
VBE2
VBE3
VBE4
VBE5
Fig. 5.16. Transistors output characteristics with breakdown voltage effects for different VBE voltages.
v) Another practical constraint is that the BJT’s current-gain drops at high frequencies. This behavior is not
observable when you analyze and simulate circuits based on the small-signal model that has been introduced in this section because this model does not include any capacitances. In reality, there are parasitic junction capacitances within the BJT device that form poles with the BJT’s internal resistances, causing current-gain roll-off at high frequencies. Figure 5.17 shows a simplified plot of β vs. frequency in logarithmic scale. The current-gain for small-signal AC calculations can be approximated as βAC, which is relatively flat over frequencies below a corner frequency fC. When using the presented small-signal model, it is important to be aware of the assumption that the BJT is operating in this frequency range. Otherwise, your hand calculations will not agree well with actual measurement results or with simulation results based on transistor models that include parasitic capacitances. As the frequency is increased, there will be a point at which the BJT does not provide current amplification anymore. This frequency is called the transition frequency fT, which is also sometimes referred to as unity-gain frequency since β = 1 at this frequency. Datasheets normally include the specifications of fT and fC, or a plot similar to Figure 5.17. Depending on the fabrication technology, fT of BJTs could as low several megahertz and as high as tenths of gigahertz.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
14
Fig. 5.17. Frequency-dependence of .
The transition frequency depends on parasitic capacitances within the BJT, but it is also proportional to the transconductance parameter gm. Since gm can be increased by biasing the transistor with a larger collector current according to equation 5.12, you can improve the frequency response of BJTs by selecting the proper bias conditions. As shown in Figure 5.18, fT is low for small values of ICQ, and it tends to increase with ICQ until the non-idealities explained in ii) begin to degrade β.
Fig. 5.18. Transition frequency vs. collector current.
V.5. Basic Amplifier Configurations and Equivalent Circuits. V.5.1. Common-emitter amplifier with the loading effects. The circuit shown in Figure 5.19a is known as common-emitter amplifier because the emitter is connected to a terminal that only has a DC voltage, which is ground in this case. When the AC analysis is carried out as shown in Fig. 5.16b, where DC voltage sources are replaced by short circuits, the emitter terminal becomes an “AC ground” even if a DC voltage source is connected to it. The DC analysis of the circuit is also performed in the same manner as the previous example by replacing AC voltage sources with short circuits and AC current sources with open circuits. A new aspect related to this loaded common-emitter amplifier is that it has a coupling capacitor CC connected between the collector and load resistor RL at the output. Recalling that the impedance of the capacitor is ZC = 1/(jωCC), we can deduce that it acts like an effective short circuit when the frequency ω is high, and like an open circuit at DC when ω equals zero. Thus, it isolates the DC voltage at the collector from the DC voltage at the output terminal which is zero here. Since CC “blocks DC components”, it is often called blocking capacitor in this kind of configuration between amplification stages. The fundamental equations to be solved for DC operating point analysis are:
th
BEQ
V
V
SCQ eII , (5.20)
and from writing the equation to relate ICQ and VCEQ from the circuit in Figure 5.19:
CCQCEQCC RIVV , (5.21)
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
15
where iC was replaced by ICQ in this DC analysis. Termed static load line, this equation is a linear relationship between ICEQ and VCEQ that can be plotted on top of the transistor’s output characteristics as in Figure 5.20a. For a given VBEQ, the plotted sets of equations dictate the operating point Q (VBEQ, ICQ, VCEQ) of the amplifier. To conduct AC analysis, the transistor must be replaced by a small-signal model while the DC voltage and current sources are set to zero. We are allowed to do so according to the superposition principle, which helps us to identify the AC response of the circuit. The small-signal model of the amplifier including the load resistor RL and the coupling capacitor CC is depicted in Figure 5.19b. This equivalent circuit can be solved using conventional circuit analysis methods. For instance, it can be shown that the output voltage is given by
beCLC
CLCmo v
CRRs
CsRRgv
)(1
. (5.22)
(a) (b)
Fig. 5.19. a) Common-emitter amplifier with load resistor and b) its small-signal equivalent circuit. The voltage transfer function in equation 5.22 has a zero at DC and a pole determined by the sum of the two resistors (RC + RL) and the coupling capacitor CC. Thus, all low-frequency signals will be attenuated under the influence of the zero until the frequency of the pole. If you cannot visualize this transfer function, then you should refer to the descriptions of plotting methods in Chapter II. For frequencies above the pole frequency, the voltage gain can be approximated from equation 5.22 with s = jω = j·∞, which results in
mLC
LC
be
o gRR
RR
v
v
. (5.23a)
Notice that the equivalent load impedance is the parallel combination of RC and RL. This intuitively makes sense at high frequencies because the capacitor has very small impedance and can be regarded as a short circuit if its impedance is significantly smaller than that of the series resistor RL. Clearly, RL reduces the gain according to equation 5.23a. Many engineers call this phenomenon “loading effect”. The AC evaluation that has led to the result in equation 5.23a gives a good indication of the amplifier’s frequency response at low and medium frequencies. However, you should not forget that the gain will drop at very high frequencies due to the β degradation described in Section V.4.5. Considering that a reduction of the AC current-gain leads to a smaller AC collector current, the impact of low β at high frequencies can be observed by rearranging equation 5.23a to write the output voltage in terms of the collector current:
cLC
LCbem
LC
LCo i
RR
RRvg
RR
RRv
. (5.23b)
The loading effect implications are evident in Figure 5.20b, where the slope of the static load line for DC analysis is -1/RC, but the AC voltage gain is defined by the dynamic load line with a slope of -1/(RC||RL) according to equation 5.23b. The load resistance RL reduces the voltage gain, which is why it is important to maximize this resistance when the goal is to achieve high gain. In case multiple amplifiers are cascaded, the resistance RL is replaced by the
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
16
input impedance of the following amplification stage. Therefore, each cascaded amplifier should be designed with high input impedance to avoid gain reductions from the loading effect on its previous stage. Notice in Figure 5.20 that the same AC base-emitter input voltage signal (shown on the right side of each plot) produces different output signal amplitudes depending on the loading conditions: The blue lines in Figure 5.20a cross the load line defined by 1/RC, which maps into the large collector-emitter output voltage swing that is shown below the x-axis. In contrast, the red dynamic load line with the influence of RL in Figure 5.20a crosses the blue lines at different points, which produces less collector-emitter output voltage swing.
iC
VCE
QICQ
VCC
VCC/RC
0.3V VCEQ
static load line (1/RC)
dynamic load line (1/[RC||RL])
VBEQ
(a) (b)
Fig. 5.20. Output characteristics and load line of the BJT in the common-emitter amplifier configuration: a) operating point Q defined by the combination of VBEQ, ICQ, and VCEQ, b) comparison of the static load line with the dynamic load line that is affected by RL, where the Q point and transistor output characteristics are unchanged.
V.5.2. Common-emitter amplifier with resistive biasing. In the simplest common-emitter amplifier, the emitter terminal is connected to ground as depicted in Fig. 5.21. The DC base-emitter voltage is determined by the resistors R1 and R2, while the input signal is AC coupled through the DC blocking capacitor CB. As a result, vB has two components: a DC voltage generated from the resistive divider connected to VCC, and an AC signal due to vi.
VCC
vi
R1
R2
vBCB
Zb
RC
voiC
CC
RL
Fig. 5.21. Schematic of the basic common-emitter amplifier:
As previously, the superposition principle can be applied to any linear system. Hence, assuming that the transistor is operated in the vicinity of the Q point, the amplifier can be approximated as a quasi-linear device for which the DC and AC modes of operation can be analyzed independently. Notice that the output signal will be composed of both DC and AC components, but the analysis of the circuit can be split to consider each signal at a time. First, let us analyze the effect of the DC signal based on the simplified diagram in Figure 5.22a, where vi = 0. Since the impedances of the capacitors CB and CC are extremely high at DC and low frequencies, they isolate the DC biasing of this circuit from the rest of the system. This has the advantage that the DC bias circuitry can be designed independently from any previous or subsequent stages on an integrated circuit, as well as the DC voltage levels associated with any external sources or loads. For this reason, the use of blocking capacitors is particularly common in multi-stage amplifiers.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
17
VCC
RCICQ
RB
VB
+
-CCB V
R
R
1
+
-VCEQ
IBQ
iBB
BB vCsR1
CsR
(a) (b)
Fig. 5.22. Simplified equivalent circuits: a) for DC analysis, b) for analysis. The DC analysis is simplified further if the base-emitter junction is modeled by a fixed DC voltage source of VBE = 0.7 V. At the input of the amplifier, the following mesh equation can be written:
VRIVR
RV BBQCC
BBB 7.0
1
. (5.24)
Furthermore, the collector current and resistor RC relate to the collector-emitter voltage as follows:
CCQCEQCC RIVV . (5.25)
Since IBQ = ICQ/ cannot be considered as an independent variable, we have two equations and four variables: R1, R2, RC, and ICQ. This gives us two degrees of freedom for the selection of these variables. However, both RC and ICQ are constrained by the AC parameter requirements as we will see shortly. It is therefore helpful to conduct an AC analysis before using equations 5.24 and 5.25 to choose the appropriate component values.
iBB
BB vCsR1
CsR
Fig. 5.23. Small-signal model for the common-emitter amplifier with resistive biasing.
AC analysis is carried out by shorting the DC voltage sources and opening the DC current sources if any are present. The resulting circuit is drawn in Figure 5.22b. As next step, the transistor can be replaced by a small-signal model. Here, the model in Figure 5.19b is employed to redraw the small-signal equivalent circuit as in Figure 5.23. Now, the analysis of this circuit is straightforward: Since the base-emitter junction is modeled by the small-signal resistor r, the voltage vbe is the fraction of the Thevenin equivalent AC input voltage (vi·sRBCB /[1 + sRBCB]) that drops across r. Resistor ris in series with the parallel combination of RB and CB, hence this voltage division can be expressed as:
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
18
iBB
BB
Bi
BB
BB
BB
Bbe v
CR||rs1
CsR
Rr
rv
CsR1
CsR
CsR1
Rr
rv
. (5.26)
This equation shows that the input circuit configuration has a high-pass transfer function due to the zero located at = 0 and the pole at = 1/(r||RB)CB. Evidently, the pole frequency is determined by the product of the blocking capacitor and RB||r. As the signal frequency increases above the pole location, the transfer function tends to approach a gain of unity because RB is typically chosen much larger than r. Hence, the base voltage becomes very similar to the input signal vi at frequencies far beyond the pole location. When designing such a resistively biased input stage, you should be careful during the selection of the component values because the capacitor CB not just blocks the DC voltage but also attenuates low-frequency signal components below the pole frequency given by 1/(r||RB)CB. Since RB is usually much bigger than r, its value has negligible impact on the pole frequency, which can typically be approximated as [(1/[1/r+ 1/RB])CB]-1 ≈ 1/rCB, where RB >> r. However, the bigger the blocking capacitor is, the lower is the pole frequency. As a result, large capacitors are often used for reducing the pole frequency in order to avoid attenuation of desired low-frequency signals. Under the condition RB >> rnotice that
BπBBπpii
Bπ
B
Bπ
πωωbe Cr
1
CR||r
1ω,vv
R||r
R
Rr
rv
p
. (5.27)
The output voltage is determined by the small-signal transconductance gm and the load impedance composed of the resistors RL and RC as well as the coupling capacitor CC. The output voltage can be found by solving the two nodal equations at each side of CC. Since the analysis is straightforward, the details are not provided here, but it is advisable that you solve the circuit and verify the result below.
iCmBBB
BB
CLC
CL
beCmCLC
CLo
vRgRr
r
CRrs
CsR
CRRs
CsR
vRgCRRs
CsRv
||11
1 (5.28)
The overall transfer function has two zeros at DC and two poles. These are generated by the blocking capacitors CB and CC, and the poles are located at the following frequencies:
BBPinputP CR||r
1
1 , (5.29)
and
CLCPoutputP CRR
1
2 . (5.30)
At high frequencies, the impedances of the blocking capacitors decrease and they can be considered as short circuits in first-order approximations. The high-frequency gain at frequencies above both poles can be approximated from equation 5.28 by substituting = ∞ and assuming RB >> r, which yields
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
19
LCmLC
LCm
i
o RRgRR
RRg
v
v||
if ω >> ωP-input, ωP-output . (5.31)
Based on the small-signal diagram in Figure 5.23, the intuitive reasoning that also leads to the above result is as follows: For frequencies such that the impedance of the blocking capacitor CB becomes very small compared to resistors R1, R2 and r, it can be considered as a short circuit. Since RB >> r, the entire input voltage appears at the base-emitter junction, and it is converted into current by the transistor’s transconductance (iL = -gmvbe ≈ -gmvi). Similarly, at the output, the blocking capacitor CC can be treated as a short circuit, leading to an effective load impedance given by RC||RL. Therefore, equation 5.31 can be easily derived from this intuitive analysis because vo = iL·(RC||RL) ≈ -gm(RC||RL)vi. The magnitude response of the circuit that shows the effect of the poles and zeros on the low-frequency response is depicted in Figure 5.24. Either ωP-input or ωP-output could be the pole at lower frequency (ωP1), which depends on the selected component values in the design.
Fig. 5.24. Typical magnitude response for the common-emitter amplifier with blocking capacitors.
To minimize the low-frequency bandwidth limitation, the blocking capacitor values must be as high as possible. If the values of CB and CC are very high, then the frequencies of the poles are very low and the frequency range with high voltage gain (i.e. bandwidth) will be widened by extending it low lower frequencies. However, high-valued capacitors tend to have larger dimensions, which usually results in space constraints and increased product cost. As an engineer, you will often have to find the optimum trade-off between performance specifications such as bandwidth, and economic factors such as the fact that it will cost more to fabricate and package a large circuit. At the same time, you also not forget the practical concerns. When designing a cellular phone for example, you might want to consult datasheets before selecting a 100 farad capacitor because it is not likely to fit into the standard housing, or will make it very difficult to find customers who are willing to buy a cellular phone that does not fit into their pockets. V.6. Transistor characterization and design approach. V.6.1. Common-emitter amplifier. Let us consider the design of an amplifier with a high-frequency gain requirement of 34 dB. The Q2N222 bipolar transistor will be used in this example because it is available as a part in most PSPICE simulator versions. Even though the model contains the small-signal parameters of the transistor, it might not always model the high-frequency current gain roll-off accurately. You can check whether the model in your simulator accounts for high-frequency effects by setting up a test circuit such as the one displayed in Figure 5.25. In this case, the BJT is biased with ideal sources having zero series resistance. Thus, when you perform an AC simulation over a wide frequency range (e.g. 0.1 Hz – 10GHz), then the change of the ratio of the AC collector current and the AC base current (ic/ib) will be solely due to transistor non-idealities if these are taken into account by the simulation model that you are using. If this ratio does not decrease at high frequencies, you should use fT and fC from the datasheet to determine up to which frequency range your simulation results are reliable based on the plot in Figure 5.17. In general, some of the small-signal parameters of this transistor are: βAC ≈ 200 = βDC ≈ 200. These parameters correspond to the transistor 2N222 offered by Philips, whose parameters were obtained through LT-SPICE; the model used in PSPICE, however, shows βAC ≈ 170 and βDC ≈ 170 which correspond to the device offered by another
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
20
vendor. These parameters can also be extracted from PSPICE simulations provided that you use the SPICE model provided by the vendor. For this example, let us assume they are both equal to 200 from previous characterization. Their typical ranges also vary among different manufacturers, but are usually listed in the datasheet. You can take these values as a reference, but should also perform your own transistor characterization as explained next. DC transistor characterization. A simple transistor characterization setup with two voltage sources is shown in Figure 5.25. The input characteristic curve can be obtained by setting the collector-emitter voltage VCEQ to a reasonable value such that the transistor operates in the linear region (VCE > 0.3 V). In this example, VCEQ is equal to 1 V. If the base-emitter DC voltage is swept from 0 up to 0.7 V or more, then the input characteristic curve in Figure 5.26 is obtained. The small-signal transconductance can be determined from this plot by finding the slope around the operating point, as marked in the figure. Please compare the value obtained from the plot and the one calculated with the previously introduced theoretical approximation: gm ≈ 40·ICQ = 48 mA/V at ICQ = 1.2 mA. During the design procedure of an amplifier, you will often have obtained a desired transconductance value from the analytical small-signal analysis and calculations. In such a situation, you can use the input characteristic plot to determine the collector current ICQ and corresponding VBEQ voltage required to bias the transistor so that it has this desired transconductance.
Fig. 5.25. Basic simulation setup for transistor characterization. Since 0.3 V < VCE = 1V and VBE = 0.75V, the
transistor is biased in the linear region.
Fig. 5.26 Input characteristic curve for the selected BJT device: IC vs. VBE, where VCEQ = 1 V for this case.
Slope = gm @ IC=1.2mA
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
21
The transistor’s output characteristic curve is obtained by setting the DC base-emitter voltage in the simulation setup such that the desired collector current is generated. Sweeping VCE while keeping VBE fixed will result in the output characteristic plot depicted in Figure 5.27. In this example case, the DC base-emitter voltage is fixed to 0.65 V, leading to a collector current around 1.2 mA. Recall that we have previously defined the boundary of the linear region with VCE > 0.3 V, but also with VCE > 0.5 V. This boundary depends on the type of BJT being used, and it also has a small dependence on VBEQ. Nonetheless, you can see from the figure that the requirement of VCE > 0.3 V for linear operation is more appropriate for the Q2N222 device.
Fig. 5.27. Transistor output characteristic curve: IC vs. VCE for VBEQ = 0.65 V.
The transistor’s output conductance go can be obtained from the output characteristic plot by finding the slope of the curve when the device is operating in the linear region. For the example shown in Figure 5.28, go is approximately 2·10-5 A/V around the operating point defined by ICQ = 1.2 mA and VCEQ = 2.5 V. The corresponding output resistance is rce= 1/ go = 50 kΩ.
Fig. 5.28. Output Characteristic curve (zoomed-in): extraction of output conductance (1/resistance) from the plot.
The input impedance is obtained by plotting IB vs. VBE as shown in the simulation result displayed in Figure 5.29, which was generated with a DC sweep of VBE while maintaining VCE at 1 V. The transistor’s input resistance r is
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
22
equal to Vth/IBQ. Since r = 1/g, the input resistance can be derived from the slope (g) of the IB-VBE curve at the proper base-emitter voltage, which is 0.65 V in this example.
Fig. 5.29. Simulated IB-VBE curve for the BJT under characterization. The input conductance g and the input
resistance r (=1/ g) can be obtained from this plot.
Other important parameters are the static current gain DC and the small-signal current gain AC. The former parameter must be used for the computation of the DC-operating point, and it is defined as the ratio of ICQ to IBQ. In the following simulation, DC is approximately equal to 200, where IBQ = 30 A is the value from the DC operating point information after a DC simulation. The dynamic small-signal current gain relates the AC collector current and the AC base current, therefore it has to be used whenever AC analysis is carried out. It can be extracted from a plot of IC vs. IB around the operating point. To do so, you can use the VBE DC sweep range in Figure 5.29, and afterward place IC on the y-axis and change the data on the x-axis from VBE to IB. Using the resulting plot (Figure 5.30) to calculate AC = IC/IB around the operating point with ICQ = 5 mA gives AC = 200 for this bias condition.
Slope = g @ IB=5A
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
23
Fig. 5.30. IC-IB curve: DC (= ICQ/IBQ) and AC (= ICQ/IBQ) can be obtained from this plot.
V.6.2. Example design procedure. A particular case of the previously discussed common-emitter amplifier is depicted in Figure 5.31, which is biased with a single resistor between the base terminal and the supply voltage. Let us use the topology in Figure 5.31 for the design on an amplifier with a required voltage gain of 54 (≈ 34.6dB). In a first approximation, the voltage gain can be inferred from equation 5.31:
thCCQCmv V/RIRg|A| 4, (5.32)
which can be rearranged to: V.mV.RI CCQ 4126853 . (5.33)
Based on this design constraint, the required output impedance is around 1KΩ if a collector current of 1.4 mA is chosen. Although the selection of the collector current is somewhat arbitrary in this example, the value of the load impedance is quite critical. It should be selected such that R4 is much larger than the input impedance of the following stage. In addition to the collector current’s impact on the gain (equation 5.32), it defines the base current together with the transistor’s value of and thereby it also affects the input resistance r. However, at this stage of the discussion we would like to focus the attention to the analysis of the DC bias conditions. With a collector current ICQ of 1.4 mA and the assumption that DC is around 200, the expected base current is IBQ = ICQ / DC = 7µA. Assuming VBEQ = 0.7 V in a first design iteration, the value of resistor R1 connected at the transistor’s base can be calculated as: R1 = (Vsupply – 0.7V) / IBQ = (5V – 0.7V) / (7·10-6) = 614 KΩ. Let us simulate the circuit in PSPICE to assess the voltages and currents of the BJT that define its DC operating point with the above selection of resistance values. The results can be obtained from the output file available in PSPICE, from which you can make several important observations:
1. The values obtained from SPICE are
V(vbase): 0.664408 voltage V(collector): 3.55373 voltage Ic(Q1): 0.00144627 device_current Ib(Q1): 7.06122e-006 device_current Ie(Q1): 0.00145333 device_current
2. Therefore, the expected voltage gain (≈ ICQ·R4/26mV) should be slightly greater (34.9 dB) than required. 3. The actual base-emitter voltage of 0.66 V is less than the 0.7 V that was assumed in our previous
calculations. Accordingly, we have to re-calculate the value for R1 that results in a better operating point.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
24
4. Since the voltage across R1 is larger than expected, the base current is larger than in the hand calculations. This is the reason for the higher collector current.
Before we recalculate the amplifier’s components, let us simulate the circuit to find out its AC response. The following results were obtained with the AC analysis option in PSPICE. The circuit has a low-frequency pole that it is defined by the capacitor C1 and the parallel combination of R1 and r according to the discussion in Section V.5.2. If this is not obvious to you by inspection, then it would be helpful for you to draw the AC equivalent circuit with a -hybrid model for the transistor, from which you can see that r and R1 of this circuit appear in parallel. The base resistance rof around 3.7 kΩ dominates the value of the input impedance because it is much smaller than R1. Thus the location of the pole is around 1/rC rad/sec (= 4.3 Hz), which is reasonably close to the pole (not visible in Fig. 5.32) in the simulated AC response. The simulated medium-band frequency gain is around 34.8 dB according to the plot in Figure 5.32. This result is not bad at all, since the ideal gain is 34.6 dB.
Fig. 5.31. Common-emitter amplifier for this simulation example.
Let us find out the reasons for the gain discrepancy between the first analytical design iteration and the simulation results. Prior to the second design iteration, it is imperative that you check the transistor’s simulation parameters. In PSPICE schematic simulation program, this can be done by analyzing the OUTPUT FILE as follows: Click the analysis tab and select the option “Examine Output”. The output file that will be opened is important because it lists the AC and DC parameters that are used by the simulator based on the device model parameters. In the second design iteration, it is advisable to use the proper DC and VBEQ from the first DC simulation. To find the value for R1 in this design example, we take into account that the collector current is defined by the base-emitter current. Therefore, the priority is to adjust the base current using the proper values for bias resistor R1, VBE, and βDC. After repeating the DC simulation with a newly selected value for R1, the collector current and collector voltage are closer to the expected DC value (VCEQ_ideal = 3.6 V). It is expected that the small-signal gain is also closer to the design target. The circuit was re-simulated in PSPICE, and the frequency response from the AC simulation is plotted in Figure 5.34. The AC gain is 34.6 dB, which is closer to the 34 dB target value. This agreement is acceptable for most practical applications, but the above design steps could be repeated if necessary.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
25
Fig. 5.32. Gain and Phase responses of the example amplifier for medium-band frequencies after the first design
iteration.
Fig. 5.33. Second DC simulation of the common-emitter amplifier: the displayed voltages for the new operating point are closer to the desired values due to the new value of bias resistor R1.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
26
Fig. 5.34. Second design iteration: gain of the example amplifier for medium-band frequencies. Here, the ideal gain
requirement is very close to the simulation result of 34.6 dB.
V.6.3 General characterization procedure for SPICE simulators. To design a good amplifier is not an easy task because it depends on many factors. Usually, there is not a universal solution, but a good engineer must be able to find an appropriate trade-off between performances and cost in terms of power, silicon area (or circuit board area), and required components. In most of the practical cases, the solution must be tailored to system specifications that reflect the priorities of customers and managers. Generally speaking, the design of amplifiers involves several fundamental steps:
i) Understand the problem and obtain the specifications for your design. Before you start your design, obtain the most important amplifier specification requirements: input impedance, output impedance, voltage/current/power gain. You also have to know or estimate the expected signal swing in various amplifier stages. This step is fundamental for the selection of components or the fabrication technology to be used because there are voltage range restrictions associated with these choices.
ii) Characterization of transistors. The component vendors provide the technical information about the devices, which is why you should familiarize yourself with the datasheets of all components that are under consideration for the design. It is also advisable to perform SPICE simulations in order to characterize the active devices using the proper models as discussed in Section V.6.1.
iii) Choose proper components and devices. This is an application-dependent selection. You must be sure that both passive and active devices operate properly for the given specifications such as maximum frequency of operation, power dissipation, supply voltages, expected signal swings, impedance matching, etc.
iv) Identification of the correct operating points for the transistors in the circuit involves the most critical design decisions. The amplifier’s overall gain, frequency response, and power dissipation are strongly affected by the operating point selection. For this reason, you should write the DC equations that relate the operating point parameters of the transistor(s), such as VBEQ and ICQ, with the components of the auxiliary bias circuitry based on conventional circuit analysis. In addition, it helps to write down the corresponding equations for small-signal parameters such as rπ and gm in order to take their dependence on the bias conditions into account. This approach has been briefly described in the previous examples, and it will be elaborated in the remainder of this chapter. During this design step, you should also ensure that the transistor operates properly when it is connected to its previous stage and its load or the following amplification stage. It happens very often that the standalone circuit works fine, but the performance of the circuit is degraded when it is
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
27
incorporated into a system because of finite impedances at its input and output terminals. Such interaction between blocks in a system can occur for both DC and AC characteristics.
v) Perform first approximations (hand calculations) to pick initial component values. At the beginning, we can make use of a very simple model for the active devices such as the one for the BJT that assumes VBEQ = 0.7 V. The primary goal at this stage of the design is to perform an approximate analysis that gives insights into the trade-offs and basic requirements to achieve the main specification targets. Sometimes, you might decide to modify your amplifier configuration or change to a different circuit topology after this first assessment. To obtain expressions that relate the key component parameters to specification targets, you can write down the fundamental AC equations for the specific amplifier circuit:
a. Input impedance
b. Output impedance
c. Small-signal gain
d. Poles and zeros (if needed to determine the bandwidth)
vi) Assess your first design in a SPICE simulator. Since the simple models for the hand calculations only give rough approximations, we have to simulate the circuit and compare the results with the predictions from hand calculations. First, make sure to check the transistor parameters for the selected operating point to identify critical deviations, then verify all your AC specifications, and finally run transient simulations to check that none of the signals show any significant distortion. In case your simulations do not agree with your calculations, think about the following issues and reconsider your initial design choices:
a. Is the selected transistor suitable for your application or are the conditions forcing it to operate outside of the ranges specified in the datasheet? Begin by finding any DC operating point discrepancies, and take a look at related parameters such as β and VBEQ.
b. Are your assumptions reasonable? In particular, you should check the impedances of the previous stage (or signal source) as well as the load (or next stage). If the design includes DC blocking capacitors, then you should check that the values are large enough to allow the AC signal components to pass, which you can check by running AC simulations to obtain the frequency response from the input to the output.
c. Where is the main design issue? Often minor mistakes such as specifying a bias current source as 1 A instead of 1 mA can be the cause of the problems. You will also be surprised how often engineers forget to specify the supply voltage or the correct input signal. After troubleshooting to make sure that none of those mishaps are the root cause of the problem, you can begin to search for aspects of your design that might require major improvements. For example, did you leave sufficient voltage headroom in your DC calculations to allow the maximum output voltage swing without clipping the signal? Do the small-signal parameter calculations contain errors? Are your expressions in step v) correctly describing the AC characteristics of the circuit, or should they be revised? Does the collector-emitter resistance rce reduce the gain by lowering the equivalent impedance at the output node? If rce is not significantly larger than the load resistance, then it should be included in the second iteration of the hand calculations. Notice that rce is listed as RO in the transistor operating point information provided in the previous subsection.
d. How can you overcome the main design drawbacks? Once the problems or deficiencies have been identified, the solution often involves redesign by selecting a better operating point. As mentioned before, your hand calculations in the second design iteration should include the transistor parameters listed in the operating point information after the first simulation. Normally, you have an improvement goal such as increasing the gain, which can be done by using a larger resistor at the collector output or by biasing the transistor with more collector current to boost the small-signal transconductance gm. Those actions unfortunately have trade-offs. For example, both the aforementioned changes lead to a reduction in the voltage margin for signal swings at the collector terminal. Furthermore, increasing the collector current will lower the value of rce, which might become a problem after the change because it affects the equivalent output impedance at the collector node. You can raise your awareness of those trade-offs by referring to the equations that you wrote in your first hand calculations.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
28
vii) Re-evaluate your specifications, assumptions, and circuit architecture. You should recalculate all amplifier components after changing one or several parameters. As you can see from the analysis in this chapter, the transistor parameters and bias circuit component values are typically interrelated in a way that one change mandates a completely new design iteration. Always start by checking the simulated VBEQ voltage of the transistor(s) and the current gains (β) for DC and AC analysis. Then, use the latest simulation values to recalculate new components.
viii) Check your final design through simulations. It is not uncommon that you need at least 2 iterations to come up with a reasonable solution. Your characterization should always cover the complete set of simulations. For instance, if you make a design change to increase the gain, it is not sufficient to run AC simulations and stop after determining that the gain meets your target. Instead, you should repeat all DC simulations and other AC simulations such as the input/output impedance assessments. An important final verification step is to run transient simulations to verify that the signal gain is achieved without clipping or other distortion.
V.7. Common-Emitter Amplifier with Emitter Degeneration Resistors. A very general common-emitter configuration is displayed in Figure 5.35. The DC operating point is more robust when an additional resistance is added between the emitter terminal and ground. As will be evident in the following discussion, this resistance (RE1 + RE2) makes the circuit less sensitive to temperature variations and it allows more control over the operation point. Without capacitor CE, a drawback of this topology would be that the resistors RE1 and RE2 both decrease the small-signal voltage gain. To alleviate this issue, the large capacitor CE eliminates the effect of RE2 on the AC voltage gain without affecting the DC operating point. Hence, CE enables more freedom to select a resistance (RE1 + RE2) for DC operating point stabilization, while RE1 becomes the emitter degeneration resistance at AC signal frequencies. In the future, you might encounter circuits in which the emitter degeneration resistance RE1 is used to improve the linearity characteristics of the amplifier. As discussed later in this section, the AC input resistance of the amplifier is increased due to added emitter resistance.
VCC
RC
vi
vC
RB1
RB2
vB
RE1
RE2 CE
CBRS
RL
vo
CL
Fig. 5.35. General common-emitter amplifier configuration.
V.7.1. DC analysis. The DC equivalent circuit of the general common-emitter amplifier is shown in Figure 5.36a. Using the same circuit analysis approach as used for the previous examples, the following set of equations can be written to relate the operating point parameters:
2121
1
17.07.0 EE
BCQEEEQBBQCC
B
BBB RR
RIVRRIVRIV
R
RV
, (5.34a)
21 || BBB RRR , (5.34b)
2121
EECCQCEQEEEQCCQCEQCC
RRRIVRRIRIVV . (5.34c)
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
29
VCC
RCRB1
RB2
vB
RE1+RE2
VCEQ
ICQ
IEQ
+
-
vi
RE1
reie
iezb RC||RL
vo
RB
ib
zi
(a) (b) Fig. 5.36. Common-emitter amplifier with source degeneration: a) DC equivalent circuit, b) small-signal model for
AC analysis. The resistance RS of the signal source is ignored in the AC analysis under the assumption that it is very small compared to the input impedance zb.
V.7.2. AC analysis. The AC analysis of the topology is a bit more complicated due to the resistor RE1 connected at the emitter of the transistor. It is easier to analyze the circuit with the T-model for the BJT, resulting in the small-signal circuit in Fig. 5.23b. For first approximate calculations, we ignore the effect of the collector-emitter resistance. To further simplify the analysis of this topology, we will consider the case with RS = 0. Even though this approximation leads to reasonable results when RS << zi, it will be apparent at the end of this subsection that the effect of RS can be easily incorporated into the final equations. To reveal one benefit of adding resistor RE1, let us determine the input impedance of this topology. Looking into the base of the amplifier, this input impedance can be shown to be
11 11)1/( EEe
e
i
b
ib RrRr
i
v
i
vz
. (5.35)
A desirable effect of the resistor RE1 is the term (1+)·RE1 in the above expression. It allows the input impedance to be boosted by adding emitter degeneration resistance. If large input impedance is required, the resistors RB1 and RB2 must be increased as well because the impedance (zi = zb||RB) seen by the input voltage source vi is the parallel combination of zb and RB. From the small-signal equivalent circuit, the amplifier’s voltage gain is derived based on the following equations:
LCeo RRiv , (5.36)
where:
i1Em
m
1Em
i
1Ee
ie v
Rg
g
Rg
v
Rr
vi
. (5.37)
Therefore, the overall voltage gain can be obtained as
1Em
LCm
i
0V
Rg
RRg
v
vA
. (5.38)
A drawback of the emitter degeneration resistor RE1 is that the voltage gain is reduced due to the term gmRE1. There is a clear design trade-off whenever we add the emitter degeneration resistor: the larger the resistor RE1 is, the larger becomes the input impedance according to equation 5.35, but the smaller is the voltage gain as can be seen in equation 5.38. This exchange is unavoidable in BJT based circuits, and the emitter resistance RE1 must be judiciously selected to satisfy the input impedance constraint without excessive degradation of the voltage gain.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
30
Another benefit of the emitter degeneration is that it reduces the circuit’s sensitivity to temperature variations. Remember that , gm, r and re are temperature-dependent parameters. At room temperature, the voltage gain can be approximated as follows:
1V 401
40A
ECQ
LCCQ
RI
RRI
, (5.39)
where we assumed that is very close to 1. If a robustness required, it is advisable to design the circuit such that 40·ICQRE1 >> 1. In practice, 40·ICQRE1 is often designed to be between 4 and 10. In that case the above expression can be further reduced, leading to the following approximation:
1E
LCV
R
RRA . (5.40)
This result shows that if the emitter degeneration resistance is sufficiently high, the voltage gain is not very sensitive to temperature-dependent transistor parameters such as r, re or gm. The voltage gain mainly depends on the ratio of the load impedance RC||RL and the emitter degeneration resistance RE1. A design example based on this circuit is discussed in Section V.10. V.8. Common-Collector Amplifier. The common-collector amplifier is frequently used for coupling circuits with insufficient driving capabilities to heavy loads. Here, “driving capability” refers to an amplifier’s ability to provide enough current to the load in order to generate the output voltage swing that is expected for the specified gain. When using this terminology, a “heavy” load is a load that requires a relatively large output current, such as a small-valued resistor. On the other hand, a “light” load only draws a small amount of current at the amplifier output. As a side comment, be cautious when using technical jargon with such a relative nature because the classifications vary among engineers who work on circuits that are intended for diverse applications, which often require different transistor fabrication technologies. For example, a designer of an integrated circuit that drives the speaker of a portable MP3 player might consider 10mA as a heavy load, while an electrical engineer designing motor control circuitry with discrete components for a manufacturing plant could regard the same current as a light load. With that said, let us get back to the common-collector amplifier, which has a voltage gain less than but close to unity. However, its current gain is around 1+. Due to this property, the amplifier can be considered as a power amplifier because it preserves almost the same voltage swing while amplifying the current. Therefore, it is often used as a so-called “buffer” for driving low load impedances such as speakers, motors, etc. The schematic of such a common-collector amplifier is presented in Figure 5.37. As you can see, the collector is connected to the supply voltage VCC that becomes an AC ground in the small-signal analysis, which is why the configuration is often named “common-collector” amplifier. As mentioned above, the output voltage at the emitter closely follows the input voltage at the base. Due to this property, some engineers prefer to call the circuit an “emitter-follower”.
VCC
vi
vE
RB1
RB2
vB
RE
CBRS
RL
vo
CL
Fig. 5.37. Common-collector configuration, where the signal is injected at the base terminal and the output is at the
emitter terminal.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
31
To conduct the DC analysis, the AC input signal source is short-circuited and the capacitors are replaced by open circuits. The resulting circuit is shown in Figure 5.38a. Resistors RB1 and RB2 together with the effect of VCC can be modeled by an equivalent circuit as depicted in Figure 5.38b, in which the equivalent voltage VBB and resistance RB are:
CCB
BCC
BB
BBB V
R
RV
RR
RV
121
2 , (5.41)
21
2111 ||
BB
BBBBB RR
RRRRR
. (5.42)
At the input, the base and emitter currents are related by the expression
CQEB
EQEBQBBB IRR
VIRVIRV
1
7.07.0 , (5.43)
and the equation at the transistor’s output yields
CQECEQEQECEQCC IRVIRVV . (5.44)
(a) (b)
Fig. 5.38. a) Common-collector equivalent circuit for DC analysis, and b) modified equivalent circuit.
The AC equivalent circuit for the amplifier in Figure 5.37 is obtained by setting VCC equal to zero and replacing the transistor by its small-signal model. It is easier to find expressions for this common-collector configuration when the T-model is used as shown in Figure 5.39a. Here, rce in the model is omitted for simplicity under the following assumption: rce >> RE||RL. Remember that the common-collector amplifier typically drives a low impedance load, which makes this a very safe assumption for many cases. Similar to the common-emitter configuration, the small-signal transfer function has two poles due to the coupling capacitors (CB, CL) and the resistors associated with them. In the medium frequency band, where the capacitors can be considered as short circuits, the equivalent circuit can be simplified as in Figure 5.39b.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
32
vi ve
RB
RE
CBRS
RL
V0
CL
reie
ieib
viRB
REIIRL
RS
v0
reie
ieibvb
zbzi
vb
(a) (b)
Fig. 5.39. Small-signal diagram for the common-collector amplifier: a) valid model for low and medium frequencies, in which the AC coupling capacitors are included; b) small-signal model valid for medium frequencies
only, where the poles introduced by the capacitors are ignored.
Based on the simplified small-signal model, the common-collector amplifier’s input impedance zb is given by
LEe
LEe
b
b
e
b
b
bb RRr
RRr
vv
iv
i
vz ||1
||
1
1
. (5.45a)
Notice that the input impedance measured at the base of the transistor depends on the overall resistance seen at the emitter terminal. In this case, the impedance looking into the base terminal is equal to the multiplication of the term re+RE1||RL with the emitter-base current gain 1+. It is interesting to point out that equation 5.45a can be expressed as
LEb RRrz ||1 , (5.45b)
where the relationship re = rπ/(1+β) was substituted to rewrite the equation. The above impedance is an important property of BJTs with emitter resistances. From inspection of the π-hybrid BJT model with grounded emitter, the impedance looking into the base is only rπ. But, the resistors attached at the emitter terminal increase the input impedance seen at the base terminal from rπ to a value of rπ + (1+)(RE1||RL). Now, the overall input impedance seen by the input voltage source vi can be identified:
LEBbBi RRrRzRz ||1|||| . (5.45)
Furthermore, the voltage vb directly at the base is generated from the input voltage source and the voltage divider involving RS and zi, for which the transfer function is
BbS
Bb
iS
i
i
b
RzR
Rz
zR
z
v
v
||
||
. (5.46)
To avoid signal attenuation in the input stage, it is desirable to design the circuit such that RS << zi = zb||RB. As aforementioned, the impedances connected at the emitter help us to this end. From Figure 5.39b, it can be noticed that the output voltage depends on vb as follows:
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
33
LEe
LE
b
o
R||Rr
R||R
v
v
. (5.47)
By combining equations 5.46 and 5.47, the overall voltage gain can be obtained as
LEe
LE
BbS
Bb
b
o
i
b
i
o
R||Rr
R||R
R||zR
R||z
v
v
v
v
v
v. (5.48)
Clearly, the amplifier’s voltage gain is less than unity because it is composed of the multiplication of two factors that we could call the input and output attenuation factors. The input voltage source impedance RS reduces the overall transfer function, and its effect is minimized if the input impedance of the amplifier is much larger then RS. The overall attenuation of this common-collector amplifier is minimized if the conditions zb||RB >> RS and RE||RL >> re = Vth/IEQ are satisfied. While it is relatively easy to satisfy the former condition in most of the practical designs, this is not the case for the latter one since RL could be very small. For instance, the RL associated with many speakers is equal to 8 Ω. Consider a case where the emitter resistance re (= 26mV/ IEQ) is also equal to 8 Ω, which implies that the emitter current is 3.25 mA. Under these conditions, equation 5.48 leads to an overall voltage gain of less than 0.5. Assuming that zb||RB >> RS and RE >> RL, what emitter current is needed if a voltage gain of 0.9 is required? To meet this target, the emitter resistance re must be equal to 0.88 Ω, leading to an emitter current requirement of roughly 30 mA. What is the emitter current requirement if you have to drive a 4 Ω speaker with a voltage gain of 0.9? The fundamental equations for the DC and AC analyses of the common-collector structure have been discussed above. It is left to you to demonstrate that the small-signal current gain of the common-collector topology is close to AC if the amplifier is properly designed. Hence, its power gain is close to AC as well. V.9. Common-Base Amplifier. Contrary to the previous topologies, the input signal in the common-base configuration is injected into the emitter while the output is at the collector terminal, as shown in Figure 5.40. The base terminal is AC-grounded through a large capacitor CB connected between the transistor’s base and ground. Since the blocking capacitor provides DC isolation so that the bias resistors RB1 and RB2 determine the base current, the control over the DC base current is maintained. The other DC blocking capacitors CI and CL are used to make the biasing of this amplifier stage independent of the previous and subsequent stages. As described before, these capacitors introduce pole-zero pairs, and their values should be increased as much as possible to push the poles to very low frequencies.
Fig. 5.40. Common-base amplifier configuration.
The DC analysis of the common-base amplifier is similar to the analysis of previous circuits. By going through the same first steps of the DC analysis described in the previous subsections, the modified equivalent circuit depicted in Figure 5.41a can be obtained, where the equivalent base voltage VBB and base resistor RB are given by
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
34
CCB
BCC
BB
BBB V
R
RV
RR
RV
121
2 , (5.49)
21
2111 ||
BB
BBBBB RR
RRRRR
. (5.50)
The base and emitter currents are related by the following expression:
CQEB
EQEBQBBB IRR
VIRVIRV
1
7.07.0 . (5.51)
Application of Kirchhoff’s voltage law to the output section of the circuit produces:
CQECCEQEQECQCCEQCC IRRVIRIRVV . (5.52)
VCC
vE
RB
vB
RE
RC
vC
VBB+
-
VCEQ
+
-
ICQ
IBQ
IEQ
viRE
reie
ieib
vb
RCIIRL
vo
zi
RSve
(a) (b)
5.41. Common-base configuration: a) DC equivalent circuit, b) small-signal equivalent circuit for AC analysis. The small-signal AC analysis can be carried out by using the -hybrid or the T-model. Since the input signal is applied to the emitter terminal, it is easier to visualize the circuit’s properties with the T-model. Assuming large DC blocking capacitors are selected so that the poles generated by CB, CI, and CL are located at very low frequencies, the effect of these poles can be ignored during the AC analysis for medium band frequencies. With this condition, the small-signal equivalent depicted in Figure 5.41b is appropriate to derive the basic equations. If your hand calculations disagree with your first simulation results, then you should take a look at the collector-emitter resistance rce in the simulated operating point information of the BJT. For simplicity and more insightful equations, the expressions below are derived under the stipulation that rce is relatively large. If the value of rce is only in the same order of magnitude as the other resistors in the AC equivalent circuit, then it is advisable to repeat the analysis without removing rce in the T-model. Let us begin the simplified analysis without rce by assessing the AC input impedance. Since the base terminal is AC-grounded, it can be seen by inspection that the input impedance is
)/(1 Ee
e
Ee
EeEe
i
ei Rr
r
Rr
Rr||Rr
i
vz
. (5.53a)
If RE is selected such that RE >> re, then the input impedance is dominated by the emitter resistance re. Under this condition and recalling that re = Vth/IEQ = Vth/ICQ =/gm, the following important approximation can be made:
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
35
CQ
th
mei I
V
grz
, if 1
ECQ
th
EmE
e
RI
V
RgR
r . (5.53b)
At room temperature, this result yields
CQ
ei Irz
40
, if 1
40
ECQRI
. (5.53c)
Therefore, the input impedance of the common-base configuration can be reduced by increasing the DC collector current. This is a very important property of the common-base configuration, and it is often exploited in transimpedance amplifiers where the input signal is in form of a current and the output signal is in form of a voltage. For instance, light-sensing photodiodes in optical communication systems convert the photonic energy into AC current that must be converted into a voltage signal for further processing. As visualized in Figure 5.42, the sensor injects current into the electrical circuit, and the most basic model for it is comprised of a current source with a finite frequency-dependent impedance Zp(f). Since VCC is grounded in the small-signal equivalent circuit, Zp(f) is effectively connected in parallel with the photodiode Dp that constitutes the signal current source. Based on the current division principle, you can deduce that the ideal interface stage to the sensor has a very low input impedance to ensure that most of the injected current flows into it rather than into the impedance Zp(f). Common-base amplifiers are well suited as transimpedance amplifiers for such applications because they can present a low impedance at the interface to the photodiode, which improves the overall efficiency of the sensor circuit by minimizing the lost current iloss.
Fig. 5.42. A transimpedance amplifier with low input impedance processes the output current of a photodiode.
Now, let us continue with the AC analysis of the equivalent circuit in Figure 5.41 for the common-base amplifier. Since the base terminal is AC-grounded, the emitter current is fully determined by the transistor’s emitter resistance and the fraction of the input voltage that appears directly at the emitter terminal: ie = -ve/re. Consequently, the collector current is also proportional to the emitter voltage, leading to the following transfer function:
e
LCLC
e
eLC
e
c
e
o
r
R||RR||R
v
iR||R
v
i
v
v
, (5.54a)
where the substitutions ic = α·ie and re = -ve/ie based on Figure 5.41b were made in the equalities. The emitter resistance depends on the DC emitter current and the thermal voltage, hence equation 5.54a can be simplified as
mLCth
CQLC
e
o gR||RV
IR||R
v
v . (5.54b)
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
36
Notice that this equation is similar to that of the common-emitter transistor, but the common-base configuration does not invert the signal since the equation does not contain a negative sign. The main difference of the two structures is the input impedance. Usually, the input impedance of the common-emitter configuration is high, especially if an additional resistor is added to the emitter terminal. In contrast, the common-base configuration has a very small input impedance. To calculate the overall voltage gain of the common-base amplifier, the resistances connected to the emitter must be taken into account because they impact the attenuation of the input signal from the source. This input attenuation factor can be added to equation 5.54a by applying the voltage division principle to the equivalent circuit in Figure 5.41b, which gives an overall voltage gain of
e
LC
iS
i
i
o
r
R||R
zR
z
v
v , where Eei Rrz || . (5.55)
The above expression for the voltage gain reveals that increasing RC and RL improves the gain. Furthermore, the combination of a low source resistance and high input impedance will maximize the achievable gain of the common-mode amplifier when it is processing the input signal in voltage mode. However, optimization of zi is an intricate design issue because re must be increased to raise zi, which in turn will reduce the second factor in equation 5.55. During the design process, you can use the numerical results from the above equations to determine the optimum trade-off for the selection of parameters under the given conditions, i.e. under the constraints by specified values such as RL, RS, or the required gain. Notice, the collector-emitter resistance rce has been neglected in the analysis. How do you expect it to affect the gain? If your intuition tells you that a low rce value will decrease the gain, then you are on the right track. Still, you should verify this assumption by deriving the equation for the gain while including rce in the T-model. To get more insights into the common-base amplifier’s practical usefulness to process current-mode input signals, let us reconsider the application in Figure 5.42 and suppose that the common-base configuration in Figure 5.41 is used as transimpedance amplifier. The simplified small-signal equivalent diagram for this situation is displayed in Figure 5.43. The equation for the output voltage in terms of the emitter current is LCeLCco RRiRRiv |||| . (5.56)
Using the current division principle at the input of the amplifier, the emitter current can be expressed as
pEe
pE
ZRr
ZRpe ii
||
|| , (5.57)
where the frequency-dependent impedance Zp(f) was replaced with a fixed impedance magnitude (Zp = |Zp(fp)|) at the specific input signal frequency fp of the current ip. Combining equations 5.56 and 5.57 to obtain the transimpedance gain results in:
pEe
pE
p
o
ZRr
ZRLCi
v RR||
|||| . (5.58)
From the above expression, it is evident that the condition re << RE||Zp is optimal for high transimpedance gain because it implies that ie ≈ -ip. When RE is selected appropriately in the design to ensure that RE >> re, then iloss can be reduced by increasing the collector current ICQ according to equation 5.53b. This benefit of the common-base amplifier is clearly demonstrated by the simple analysis. Nevertheless, you will encounter limitations during the design for high transimpedance gain, which ideally requires you to increase the resistance values of RC, RL, and RE while increasing ICQ in order to reduce re. These design objectives will create voltage headroom constraints because they will lead to decreased VCEQ in Figure 5.41a, which will reduce the maximum output voltage signal swing that is permissible while guaranteeing that the BJT remains in the active region of operation.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
37
Fig. 5.43. Simplified AC equivalent circuit for the common-base amplifier with a current-mode input.
Table 5.1 summarizes the key equations for the three voltage amplifier topologies with single BJTs that have been discussed thus far. If you want to express the voltage gains in terms of the BJT’s transconductance, then you can substitute re = α/gm into the equations. This is often helpful when you design because one of the first steps is to characterize the transistor to determine its operating parameters (including gm) for various DC bias conditions.
Parameter Common-Emitter
(Figure 5.35) Common-Collector
(Figure 5.37) Common-Base (Figure 5.40)
DC Equations
21
17.0 EE
BCQBB RR
RIVV
21 EECCQCEQCC
RRRIVV
CQEB
BB IRR
VV
1
7.0
CQECEQCC IRVV
CQEB
BB IRR
VV
1
7.0
CQECCEQCC IRRVV
Input Impedance
BEi RRrz ||)1( 1 BLEi RRRrz ||)||)(1( Eei Rrz ||
Voltage Gain
1
VAEe
LC
iS
i
Rr
RR
zR
z
LEe
LE
iS
i
RRr
RR
zR
z
||
||AV
e
LC
iS
i
r
RR
zR
z ||AV
Table 5.1 Fundamental equations for the common-emitter, common-collector, and common-base configurations.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
38
V.10. Design Considerations for BJT-Based Amplifiers: Graphical methodology. . V.10.1. Design of a high gain amplifier using a common-emitter structure with limited input impedance: A graphical approach. A major concern of the analog circuit designer is to determine the appropriate transistor device parameters and circuit component values to meet a set of given specifications. This part of the design is often obscure, which is the reason why new designers sometimes think that analog design is a kind of black art. In fact, selecting proper operating points for transistors is not a trivial task because many parameters and constraints are involved. However, the simple graphical design approach discussed in this section can help us to understand the main trade-offs so that reasonable solutions can be found. A couple of design examples will be used to introduce this approach and its usefulness to you. The first example deals with the use of the basic common-emitter configuration shown in Figure 5.44a for the design of an amplifier with a voltage gain of -20 V/V that has to drive a load impedance of 10 kΩ. The limitations due to signal swing must also be incorporated as design constraints. Usually, those conditions impose some boundary conditions to the possible solution. In additions to the equations listed in Table 5.1, the following design constraints are applicable to this design: i) Expected maximum amplitude Vomax of the output signal (often called the “maximum output swing”) ii) Required voltage gain AV iii) Input and load impedance requirements iv) Restricted supply voltage VCC
VCC
vbe
RC
vo
VCE
CC
RL
+
-
ICQRC
+
-
+
-VBB
Vomax
Vomax
VCC
ICQRC
VCEmin
VCEQ-VCEmin
VCEQ
(a) (b)
Fig. 5.44. basic common-emitter amplifier: a) circuit schematic; b) Conceptual schematic showing the DC and AC components at base and emitter.
Although the input impedance is not considered in this case, it will be incorporated into the design equations in the next subsection. The following equations define the operating point of the amplifier based on the DC analysis at its output. As described in previous subsections and according to Table 5.1, the DC equation involving the collector-emitter voltage is:
CCQCEQCC RIVV . (5.59)
It is very important to consider that the DC collector-emitter voltage VCEQ must be large enough to tolerate the maximum voltage swing of the AC signal. This constraint is visualized in Figure 5.44b. The condition to ensure that the transistor operates in the linear region can be expressed as VCEmin = VCEQ – Vopk > ~300 mV. To leave room for potential temperature variations and component tolerances, let us design for a minimum collector-emitter voltage of VCEmin = 500 mV. With this safety margin, the VCEQ voltage must be chosen such that the collector-emitter voltage is greater or equal than 500 mV for the most negative signal swing at the amplifier output. This condition can be substituted into equation 5.59 to obtain the following design constraint:
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
39
C
opkCC
C
minCECCCQ R
V.VV
R
VVI
50
. (5.60)
On the other hand, VCE should not be too close to VCC in order to provide room for the positive signal swing. Otherwise, the signal would be clipped when it reaches the supply voltage level. The DC voltage drop (ICQ·RC in Figure 5.44b) across the resistor RC at the collector must be greater than Vopk, leading to the following restriction for the collector current:
C
opkCQ R
VI . (5.61)
Another important design parameter is the small-signal voltage gain from the transistor’s base to the amplifier output. From our previous analysis, this voltage gain is given by
th
LCCQLCm
be
oV V
RRIRRg
v
vA , (5.62a)
which can be rearranged to:
LC
LCthVCQ RR
RRVAI . (5.62b)
These equations can be plotted in the IC-RC plane as shown in Figure 5.45. Equations 5.60 and 5.61 represent the boundaries for the acceptable solutions, and the combination of IC and RC for the selected operating point must fall within these two boundaries to allow sufficient signal swing (Vomax) at the output under a given supply voltage limitation (VCC). Equation 5.62b gives the solutions for the amplifier’s voltage gain. These plots depend on the required voltage gain Av and the load impedance RL in addition to IC and RC. In Figure 5.45, equation 5.62b is plotted for four different amplifier gains (AV = -10, -20, -40, and -100). Notice that the boundaries (equations 5.60 and 5.61) are independent of the desired voltage gain. However, they do depend on the maximum voltage swing expected at the amplifier’s output. When only the input signal swing is specified (fixed), then you should take into account that the maximum output voltage swing and the boundaries will vary as you plot equation 5.62b for different voltage gains.
1E+3 1E+4 1E+5
Rc (Ohms)
0.01
0.10
1.00
10.00
Ic(m
A)
AV = -100
= -40
= -20
= -10
Acceptable resistance
Acceptable current
eqn. 5.62b (AV = -20)
eqn. 5.60
eqn. 5.61
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
40
Fig. 5.45. Plots of the main design equations in the IC-RC plane to determine the ranges with acceptable combinations of IC and RC for a specified voltage gain.
The results shown in Figure 5.45 are generated with VCC = 5V, Vomax = 1 V, Vth = 26 mV, RL = 10KΩ. Notice in this plot that the acceptable region is more limited for large voltage gains. For example, the solution of equation 5.49b with Av = -100 does not intersect the acceptable region. Hence, we cannot design the amplifier with a voltage gain of -100 for the given supply voltage and the desired output voltage swing. For the case of Av = -20, we can achieve the specifications by satisfying all design constraints if and only if the collector resistance is in the range of 9.5k to 58 k while the quiescent collector current is in the range of 60A-100A. Another advantage of using this graphical approach is that the power consumption can be easily optimized by selecting the largest possible collector resistor RC. For practical applications, however, it is always good practice to design the amplifier with some margin in order to accommodate both temperature variations and tolerances of the devices and components. This can be done by selecting a combination of IC and RC values on the desired voltage gain-dependent curve (equation 5.62b) that falls closer to the middle of the two boundaries (equations 5.60 and 5.61). Once the output stage of the circuit is defined and the collector current is determined, the base current can be calculated. The values of the resistors attached to the base terminal can be computed based on the needed base current and the input impedance requirement. There are two basic equations (see Table 5.1, common-emitter configuration with RE1 = RE2 = 0) and two unknown variables. Thus, we do not have any freedom for optimization when solving these equations, which are repeated here for convenience. From the DC analysis, we have
VRIVR
RBBCC
B 7.01
, (5.63)
and the amplifier’s input impedance from the AC analysis is given by
B
B
BBin
R
rr
Rr
RrRrZ
1. (5.64)
If RB >> r, the effect of the bias network on the input impedance becomes negligible so that the input impedance is dominated by the transistor’s base-emitter resistance r. For many applications it is common practice to increase the input impedance as much as possible. Thus, selecting a high-beta transistor helps because a high value reduces the base current for a given collector current requirement, which increases the input impedance of the amplifier (r= Vth/IBQ). Nevertheless, most high- BJTs cannot tolerate large currents and are therefore used as high-gain preamplifiers to avoid damage. Such a high-gain amplifier can then be connected to a low-gain buffer stage that can provide larger current to the load device. Typically, the transistors used as power amplifiers have limited values, which could be well below 50. From the above discussion, the following important interrelation is evident: By selecting the smallest possible collector current, the required base current can be minimized, which will increase the input impedance and reduce the power consumption at the same time. V.10.2. Design examples. Example 1. To design the amplifier in Figure 5.44 with a gain of -20 V/V and Vomax = 1 Vpk, we select IC = 80 A and RC = 11 k based on the plots in Figure 5.45. For this collector current, the beta of the 2N2222 transistor is close to 200, and therefore the DC base current is around 0.4 A. The base-emitter resistance ris around 49 k. Therefore, RB can be selected should be at least 300 k to maintain an overall input impedance around 40 k. Let us use RB = 400 k for this exmple, then VB = VBEQ + IBRB = 0.6V + IBRB = 0.81 V. In this hand calculation, the estimate VBEQ = 0.6V was used rather than VBEQ = 0.7V because the base current is very small. Anyway, the actual base-emiter voltage and other operating point parameters should be obtained with SPICE simulations, and the resistors should be adjusted accordingly in order to meet the collector current requirement. With some basic algebra, the value of the resistors connected to the base terminal can be determined from the chosen value of RB = RB1||RB2. In this example, RB1 could be picked as 2.2 Mand RB2 calculated correspondingly as 490kto achieve the desired
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
41
DC bias point and input impedance. When you are planning to build a circuit in the lab, be aware that your choice of resistor values will be restricted to those that you can find in the lists of standard component values. Figure 5.46 shows the voltage waveform at the transistor’s collector from a SPICE simulation of the amplifier with the selected component values with a 5 mVpk input signal swing.
Fig. 5.46. Transient output voltage at the collector terminal of the BJT. The input signal is 5 mVpk and the voltage gain is 23 V/V.
The simulated voltage gain is 15 % larger than expected. Analyzing the PSPICE output file, we discovered that the DC is 15% higher than the value used in the hand calculations. As a result, the DC collector current is also 15% larger, which is evident in Fig. 5.46, where the DC voltage component (average of the transient waveform) is around 3.7 V instead of the expected 4.1 V (= VCC - RC·IC). If needed, the base current and resistors at base can be recalculated using the obtained from the SPICE output file. Although large volatge gain can be obtained with this amplifier design, the input signal must be limited to 10 mV. Otherwise, the small-signal approximations are no longer valid. There are two main limitations with this amplifier topology. First, the input signal appears directly across the base-emitter terminals, limiting the allowed amplitude of the input signals for linear operation. Second, the input impedance is determined by the base-emitter resistance, which is not very high. To demonstrate the linearity performance limit, a 20 mV sinusoidal input signal was applied to the same circuit, for which the relevant SPICE simulation results are shown in Figures 5.47 and 5.48. Notice that the transient ouput waveform deviates from an ideal sinusoid. A Fourier analysis of this waveform can be carried out with the option in PSPICE that is available after running the transient simulation. The result in Figure 5.48 reveals that the fundamental signal component at 10 KHz has an amplitude of roughly 500 mV, while the second harmonic component at 20 KHz is 80 mV. The second-order harmonic distortion of rougly 16% is too much for most practical applications. Equation 5.10a predicts HD2 20%, which is relativelty close to the simulation results. The smaller third-order harmonic component of 10 mV at 30 KHz can also be seen in the figure, leading to HD3 = 2% in agreement with the expected 2.5% based on equation 5.10b.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
42
Fig. 5.47. Transient output waveform for an input signal of 20 mVpk.
Fig. 5.48. Frequency spectrum with the fundamental as well as second-order and third-order harmonic distortion
components of the output voltage waveform shown in Fig. 5.47. Example 2. Design of a common-emitter amplifier with larger input impedance but limited voltage gain. In many situations, the amplifier’s input impedance is also an important design parameter. The AC input impedance can be increased by adding an emitter resistance as shown in Fig. 5.49. The DC and AC equations for this scenario are given in Table 5.1, where RE1 = RE and RE2 = 0.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
43
VCC
RC
vi
R1
R2
vb
RE
CBRS
RL
vo
CL
5.49. Common-emitter amplifier with emitter degeneration resistor.
In the subsequent analysis it will be assumed that RB >> ·RE >> r. Hence, the expression for the input impedance of this topology can be simplified with this approximation:
E
CQ
thBEei R
I
VRRrZ 11 . (5.65a)
or
CQ
thiE I
VZR
1
. (5.65b)
With a high RB value, the overall input impedance only increases slightly. In addition to 5.65b, the set of equations that describe this amplifier’s DC and AC operation are revisited and rearranged below to acquire more design insights.
i
C
VcpkCC
EC
epkcpkCC
EC
CECCCQ Z
R
VA
VV
RR
VVVV
RR
VVI
5.01
15.0
min ; (5.66)
where substitutions with equation 5.65 and α = 1 were made; it is also assumed that the voltage swing at the emitter is equalt to the swing of the collector divided by the amplifier’s gain. As before, the condition VCEmin > Vcepk + 0.5V guarantees that the BJT stays in linear region with maximum signal swing. To avoid signal clipping for positive output swing, the collector voltage has to be properly selected. Since the circuit must be able to handle output voltage variations of at least Vomax, the following condition applies as well:
C
opkCQ R
VI . (5.67)
From the small-signal analysis of the circuit, the amplifier’s voltage gain can be expressed as
Eth
CQ
LCth
CQ
Ee
LC
b
o
RV
I
RRV
I
Rr
RR
v
v
1
. (5.68)
Notice that the resistor attached to the emitter reduces the overall voltage gain. The amount of gain reduction depends on the ratio of the voltage drop (ICQRE) across RE and the thermal voltage Vth The above expression does not include resistance RS of the input voltage source, but this can be easily incorporated by accounting for the voltage division associated with Zi and RS; the analysis yields
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
44
1S
Ee
LC
Ee
LC
Si
i
i
b
b
o
i
o
RRr
RR
Rr
RR
RZ
Z
v
v
v
v
v
v. (5.69)
This equation is very helpful for the selection of ICQ because it is a function of the specified AC voltage gain, load impedances RC||RL, and the input impedance requirement. Eqn 5.69 is function of RC and RE. RE could be computed from 5.65 if the input impedance is provided. The input impedance specification is usually given in form of a bound; e.g. Zi > a given value. Thus, an infinite number of solutions might be possible for 5.69. It is better practice to use this degree of freedom to guarantee the linear operation of the amplifier. Linear operation requires limiting the base-emitter voltage under a value that maintains the harmonic distortion components under the specifications. Detailed equations for the 2nd and 3rd harmonic distortions are given in the appendix; see also eqn 5.10. It is enough for his example to analyze the non-linear exponential term in equation A5.1. The non-linearity arises from the fact that the small signal drain current is generated according to the expression
..V
v
V
v
V
ve
th
be
th
be
th
beV
v
th
be
32
6
1
2
11 (570)
Where the most relevant harmonic distortion is lumped to the quadratic term; in advanced applications, the design is usually fully differential then the second harmonic distortion becomes irrelevant, then the 3rd harmonic component is the most relevant one. The second harmonic distortion is small (<1/16) if we limit the base-emitter voltage to
41 /V
v
th
pkbe ; e.g. peak value of the base-emitter voltage should be limited to 6mVpk. To guarantee this condition,
the relationship between base-emitter and input voltage can be found as follows
1S
Ee
e
Ee
e
Si
i
pki
pkbe
RRr
r
Rr
r
RZ
Z
v
v (5.71)
Then, from 5.69 and 5.71, the voltage gain becomes
pki
pkbe
th
LCCQ
pki
pkbe
e
LC
pki
pko
v
v
V
RRI
v
v
r
RR
v
v
. (5.72a)
This equation can be more practical if expressed in the following format:
vpkbe
pki
LC
thCQ A
v
v
RR
VI
. (5.72b)
Thus, designing a low-distortion amplifier requires to know-a-priori the peak value of the input signal. It is highly desirable to limit the swing of the base-emitter signal to be under 6mVpk to maintain good linearity. To complete the design procedure, Eqns 5.65 and 5.71 can be combined in a simplified result that relates the collector current and the input impedance requirement, leading to
1pkbe
pki
i
thCQ v
v
Z
VI
. (5.73)
Equations 5.66-5.72b and 5.73 can be plotted in the IC-RC plane as in Figure 5.50. To generate these plots, the following parameters and design constraints were used: = 200, Zi (equation 5.65) 150 k, RL = 10 k, Vomax = 1 Vpk, and VCC = 5V. In Figure 5.50, equations 5.72b is plotted for two different amplifier gains (AV = -5 and -10). Notice that the acceptable solution region is significantly more limited than the one discussed in the previous example. In fact, the optimal solution is determined by the intersection of curves corresponding to equation 5.72b (which determines the voltage gain) and equation 5.73 (which guaranties input impedance value), but inside of the acceptable region bounded by equation 5.66. In this case, there are no acceptable solutions for |Av| = 10.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
45
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2000 4000 6000 8000 10000 12000
Collector current (m
A)
Collector resistance RC
Equation5.66; VCC=10
Equation5.66; VCC=5
Equation5.72b; |AV|=10
Equation5.72b; |AV|=5
Equation5.73; Zi=150kΩ
Fig. 5.50. Design example 2: plots of the main design equations in the IC-RC plane to determine the ranges with
acceptable combinations of IC and RC for a specified voltage gain.
For the optimal solution annotated in Figure 5.50, the collector current ICQ is roughly 350 A and the collector resistance is around 6 k. The circuit has been simulated in PSPICE with the component values displayed in Figure 5.51. Resistor values for RB1 and RB2 were selected to be very large such that RB = 600 kΩ in order to avoid degrading the input impedance, since impedance since from the base is around 150 kΩ. Before you run AC and transient simulations, inspect the DC operating point of the circuit to verify the VCE voltage and the collector current values. The transient response for a sinewave input signal with an amplitude of 100 mVpk is plotted in Figure 5.52. The AC signal at the base terminal of the transistor is superimposed on the DC base voltage close to 0.9 V, while the output signal is centered around 2.75 V. The amplitude of the transient signal directly at the base is approximately 100 m Vpk. The output voltage amplitude is 0.5Vpk. Hence, the overall voltage gain of the amplifier is very close to the target -5V/V. The frequency response can also be obtained from PSPICE simulations using the AC analysis option. Although the magnitude and phase repsonses are not shown, we encourage you to obtain these plots to find the amplifier’s bandwidth (-3 dB frequency), which is in the range of 100 MHz. In comparison, the bandwidth of the high-gain amplifier design example discussed in Section V.6 is around 30 MHz. This demonstrates another common design trade-off: the bandwidth tends to be smaller for high-gain amplifiiers than for low-gain amplifiers with comparable devices and power consumption.
Acceptable solution
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
46
Fig. 5.51. Common-emitter amplifier with emitter degeneration resistor RE for improved linearity performance and
higher input impedance.
Fig. 5.52. PSPICE transient waveforms for a 200 mVpk input signal, revealing the voltage gain of -4.6 V/V.
The linearity of the amplifier with emitter degeneration can also be evaluated with the PSPICE transient simulation option. The first three spectral components are depicted in Figure 5.53 using the Fast Fourier Transform (FFT) tool. It can be seen that the fundamental component is close to -10dB while the second harmonic component is approximately at -50dB, which calculates to an HD2 around -40 dB. The third harmonic component is around -60 dB. An evident question is how the amplitude of the output voltage in Fig. 5.52 is correlated with the -9dB shown in Fig. 5.53? The FFT plot shows the RMS value of the output voltage; Vcollector-RMS=0.5*.21/2, which indeed corresponds to -9dB.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
47
Fig. 5.53. Spectral analysis of the output voltage at the collector. HD2 ≈ -41 dB and HD3 ≈ -62 dB.
Although the input voltage amplitude is relatively high, an important benefit of the emitter degeneration is that the swing of the base-emitter voltage reduces significantly due to the voltage swing present at the emitter. If the input signal increases such that the base-emitter voltage exceeds 10mV, the harmonic distortion components become more prominent. In figure 5.54, the harmonic distortion components for the case Vin=200mVpk are shown.
Fig. 5.54. Spectral analysis of the output voltage at the collector for the case Vin=200mVpk. HD2 ≈ -34 dB and
HD3 ≈ - 48 dB. Let us consider the small-signal equivalent circuit, which is similar to the one shown in Figure 5.36b with the exception that the signal source has a series resistance RS and that RE1 = RE. The AC base-emitter voltage can be expressed as
iEe
eb
Ee
ebe v
Rr
rv
Rr
rv
. (5.74)
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
48
With the components and bias conditions in this example design, the emitter resistance re is 74 Ω. Therefore, assuming linear operation, the base-emitter voltage vbe is attenuated to
iibe v.vv
0980Ω7475Ω6
675Ω. (5.75)
For an input voltage vi of 100 mVpk, the effective base-emitter voltage is around 10mVpk. This simple analysis helps us to understand the fundamental concept behind the linearization effect of the emitter resistor: RE effectively attenuates the base-emitter voltage swing then making the circuit more linear. Nevertheless, the accurate prediction of the harmonic distortion components based on the vbe reduction is more complicated. Some basic equations can be found in the appendix at the end of this chapter. Before concluding this section, let us have a look on the base and base-emitter waveforms for the two cases: vin-
pk=100mV and vin-pk=200mV. The signal swing at the base is exactly equal to the input signal swing since we are not considering the source resistance RS. Notice that both base voltage and base-emitter voltages look linear in Fig. 5.55a; base-emitter voltage swing is less than 10mVpk. This is not the case in Fig. 5.55b where the amplitude of the input signal is 200mVpk; base-emitter signal swing is over 20mVpk, making circuit non-linearities evident. The theoretical harmonic distortion values agree well with the simulation results.
(a)
(b)
Fig. 5.55. Base and Base-Emitter voltage measured for the case: a) vin-pk=100mV and b) vin-pk=200mV.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
49
V.11. Multi-Stage Amplifiers. An effective way to achieve high gain is to cascade multiple amplifiers as depicted in Figure 5.56. In many real world applications, the sensed signal is very small but the the output power supplied to the load is high. For example, this situation occurs in a control system in which the sensor is a wheel speed sensor and the actuator is an electrical motor that spins the wheel. If you are already familiar with control systems, then you might recall that the amplification path in Figure 5.56 is often part of a negative feedback system for improved controllability. In any case, the circuit topology of the first stage in the cascade can be selected and optimized to provide the required interface impedance for the given signal source, which is a sensor here. For instance, a common-emitter configuration with emitter degeneration resistor would be an appropriate choice if high input impedance is required. The cascade of several amplification stages allows to boost the overall gain, while the final power amplification stage is designed specifically to drive the load. As explained in Section V.8, the load driving capability is mainly determined by the amount of current that the amplifier can supply to a load at the required output voltage. Since the actuator converts electrical engergy to mechanical energy, the power amplifier requirement directly depends on the mass and rotation speed of the wheel in this case. Therefore, the amplifier in the final stage must be selected and designed according the specific requirements. If high current and power gains are required for instance, then the common-collector configuration is the best suitable topology among those discussed in this chapter.
Fig. 5.56. A cascade of amplifiers that conditions a sensor output signal and drives a high-power actuator.
The coupling between cascaded stages is an important design consideration. Consider the case in which the amplifier in Figure 5.57a comprises the first stage, which has blocking capacitors CB and CL to isolates its DC bias conditions from the sensor and the next stage. At AC signal frequencies however, the small-signal equivalent circuit (Figure 5.57b) of the amplifier in stage 1 must be carefully considered. First of all, zi >> RS should be satisfied to prevent undesired input signal attenuation. Second, zo of the first stage should be much smaller than the input impedance zi2 of the second stage to avoid coupling loss. Recall that the resistor RC at the collector in the first stage is selected as part of the design with a specified voltage gain requirement. However, when a second amplifier stage is connected, then the equivalent small-signal load impedance becomes RC||zi2. The condition zi2 >> RC ensures that this new load impedance is only slighly smaller than RC so that connection of the second stage has negligible impact on the voltage gain in the first stage. Similarly, the interface impedances between the following amplificaiton stages must be compatible by design for minimal gain redcutions due to coupling.
(a) (b)
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
50
(c)
Fig. 5.57. First amplifier in the cascade: a) schematic, b) small-signal equivalent circuit, c) macromodel. A convenient system-level design approach is to represent amplifiers as macromodels that contain the most significant specification parameters. Figure 5.57c shows a general small-signal macromodel for a voltage amplifier with an open-loop (unloaded) gain Av, input impedance zi, and output impedance zo. Notice that the gain is simply modeled by an ideal voltage-controlled voltage source, which is available in almost all circuit simulators. Thus, the macromodels can be used to quicly determine the specifications of all amplifier blocks by simulating the complete cascade for different sets of block-level specifications. Once all system-level specifications are met and the block-level requirements are identified, each amplifier block can be designed on the transistor level. This approach is more effiecient than attempting to redesign the blocks multiple times just to try out different parameters for interface impedances or voltage gains. When designing with macromodels, you have to be careful that the model parameters are achievable. For example, you should not assume zi = 100 MΩ unless you know that you can design an amplifier with this input impedance. You should also go back to the macromodel simulation setup after designing and simulating your individual amplifier blocks in order to update your macromodels with the practilly achievable parameters. At this time, you might have to go through one or more additional design itereations with the updated macromodels. Finally, you have to verify that the cascaded circuit-level designs achieve the system specifications by simulating them together to account for more realistic coupling losses than the simplified macromodel. That said, let us now derive the parameters of the macromodel in Figure 5.57c. In the following analysis we assume that RB >> ·RE >> r. Performing the small-signal analysis of Figure 5.57b, the input impedance of this amplifier topology is determined from the AC test input voltage at the base vx = vb and the resulting AC current ix that flows when the load resistor is not connected:
BEBEe
Rx
bi RRrR1Rr
i
vz
L
. (5.76)
The unloaded amplifier’s small-signal voltage gain is given by the following expression:
Eth
CQ
Cth
CQ
Ee
C
Rb
o
RV
I
RV
I
Rr
R
v
v
L
1
A V
.
(5.77)
To find the output impedance in the macromodel, the AC input is grounded (vx = vb = 0) in the small-signal analysis. Under this condition, an AC voltage applied at the output of the amplifier divided by the corresponding AC current gives an output impedance of
Cvo
oo R
i
vz
b
0
. (5.78)
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
51
Cascading the macromodels of the amplifiers in each stage helps to assess AC coupling effects based on the input and ouput impedances. This can be done by simulating or analyizing the simplified circuit such as the example two-stage amplifier in Figure 5.56. The overall voltage gain transfer function of this amplifier is
L2o
L2v
2i1o
2i1v
1iS
1ivT
Rz
RA
zz
zA
zR
zA . (5.79)
You can verify this expression by identifying the voltage divisions at the input and output of each gain stage. In summary, the equation shows the benefit of determining the input and output impedances of the standalone stages as well as their open-loop voltage gains in order to predict the overall gain of cascaded amplifiers. As in all cases, verification of the AC circuit parameters is necessary but insufficient to ensure proper operation. Since the small-signal macromodel and circuit-level AC simulations are idealizations that assume small signal variations, you should verify that your cascaded amplifier design meets the specified performance parameters during transient simulations with the maximum input voltage swing, for which the same procedure as in Section V.10 can be followed.
Fig. 5.56. Macromodel of a two-stage amplifier.
Figure 5.57 displays the PSPICE schematic of a multi-stage amplifier. From the annotated DC voltage, you can observe that the DC bias circuits for each of the four stages are independent. Thanks to the blocking capacitors C1, C2, C3, C4 and C5, the DC bias circuits can be designed individually with the equations provided in the previous sections. Various nodes of interest are labeld in Figure 5.57 and the transient voltages at these nodes are plotted in Figure 5.58. As expected, the DC voltage levels at these nodes are different, and the amplitude of the signal swing increases at each stage with the exception of the common-collector output stage in which the voltage gain is slightly less than unity.
Fig. 5.57. PSPICE schematic of a four-stage amplifier.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
52
Fig. 5.58. Transient waveforms showing the voltages at critical nodes within the four-stage amplifier.
In this example design, an ideal DC current source is used in the common-collector output stage in Figure 5.57. This bias current could also be provided with a resistor or transistor-based current source at the emitter, which would lead to some gain reduction in that stage because any finite impedance at the emitter appears in parallel with the load resistor in the AC equivalent circuit. Since all loading and interstage coupling effects are observable in transient simulations, you should always perform them to verify that the amplifer cascade achieves the desired gain with the maximum output voltage swing. In the same manner as for single-stage amplifiers, the linearity with the maximum output voltage swing of the multi-stage amplifier has to be evaluated as well. Figure 5.59 presents the frequency spectrum of the amplifier under investigation, which has HD2 and HD3 around -39dB and -45dB, respectively.
Fig. 5.59. Frequency spectra of the waveforms at the collector of Q2 (VC2) and the amplifier output (Vout). The overall output voltage at the fundamental frequency of 1 KHz is close to 2.8 Vpk. The harmonic distortions are:
HD2 30mV/2.8V = -39.4 dB, and HD3 15mV/2.8V = -45.4 dB. V.12 PNP Transistors.
VB1 VC1 VC2 VC3 Vout
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
53
V.12.1. PNP device modeling. A complementary device to the NPN BJT in Figure 5.1 can be fabricated by interchanging the doping regions of the transistor. Such a so-called PNP BJT is visualized in Figure 5.60 together with its schematic symbol. In this transistor the emitter and collector regions are composed of p-type silicon. Consequently, the current flow is predominantly due to the movement of holes. Notice that the conduction of large currents in the PNP requires different relative voltages between the emitter-base and emitter-collector junctions than in the NPN case. Furthermore, the currents of the PNP transistor flow in opposite directions compared to the NPN transistor.
(a) (b)
Fig. 5.60. a) PNP transistor and b) its symbol. Recall that in the NPN BJT the emitter current exponentially depends on the base-emitter voltage. In contrast, the emitter current of the PNP BJT is an exponential function of the emitter-base voltage:
1iE
th
EB
V
v
SE eI . (5.80)
As before, ISE is the saturation current for the emitter terminal, is a fitting factor, and Vth is the thermal voltage proportional to the temperature. The only difference between the v-i relationship of the emitter current in the NPN and PNP cases is the polarity reversal of the base and emitter voltage references. Since the relationships between the base, emitter, and collector currents in the PNP device are the same as in the NPN device, equations 5.4 – 5.6 are also valid for PNP transistors. The biasing and design procedures presented in this chapter can be used for PNP transistors as well. The only significant difference is that the DC bias conditions for the regions of operation are different. A PNP BJT is in cutoff region when VEB is less than 0.7 V. If VEB > 0.7 V and VEC < 0.3 V, then it is biased in the saturation region. To operate the PNP in the active region requires that VEB > 0.7 V and VEC > 0.3 V. As in the case of NPN transistors, the voltage ranges for the PNP’s operating regions depend on the device materials, fabrication characteristics, and manufacturer of a given transistor. AC analysis of PNP-based circuits can be performed by using the small-signal models in Figure 5.14 without a change in the polarities. The -hybrid model and T-model are both identical for NPN and PNP BJTs, and the same equations apply for the calculation of their parameters. However, the movement of holes across the depletion layer tends to be slower than that of electrons. As a result, PNP BJTs typically have less transconductance (gm) and a lower transition frequency (fT) compared to NPN BJTs when their fabrication dimensions and DC bias currents are similar. Figure 5.61 shows the schematic of a PNP-based common-emitter amplifier and its small-signal equivalent circuit with the -hybrid model. By inspecting the two diagrams, you can observe that the usage of the small-signal model is exactly the same as in the previous examples with NPN transistors.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
54
(a) (b)
Fig. 5.61. a) A common-emitter amplifier with collector and load resistors, b) its small-signal equivalent circuit employing the -hybrid transistor model.
V.12.2. Circuit examples with PNP transistors. At this point, you might ask yourself what the benefit is of having a PNP transistor that is a dual of the NPN transistor but with reduced transconductance and worse high-frequency performance. If you decide to study advanced microelectronic circuits, then you will discover that PNP are often used in voltage reference circuits, temperature sensors, and biasing circuitry because they are often easier to fabricate with conventional integrated circuit technologies. One advantage of having the availability of complementary transistors becomes evident from inspecting the amplifier in Figure 5.62, which is named push-pull follower. This circuit is very useful as a buffer when the input signal swing covers the complete span between a positive supply voltage VCC and negative supply voltage -VSS = -VCC. In this configuration, both transistors operate as common-collector (emitter-follower) amplifiers, having a voltage gain slightly less than one and a current gain around 1+. Recall that this property allows driving a low impedance load, which is why the common-collector stages are commonly found as last stage of multi-stage amplifiers. The study of the amplifiers in this chapter with single NPN devices revealed that they all have input and output voltage swing limitations which prevent signal swings from the positive supply voltage all the way to the negative supply (or ground). But such a swing can be achieved with complementary PNP and NPN devices. In the push-pull configuration for instance, the NPN transistor conducts during positive half cycles of the input signal, while the PNP transistor processes the signal during negative half cycles. Thus, the output can be linear for the input signal swing from VCC to -VSS. However, there is “dead band” when the input signal is between 0.7V and -0.7V and both transistors are in the cutoff region because the input voltage is too small. This causes distortion known as “crossover distortion” that is visible in the output signal waveform.
Fig. 5.62. A push-pull follower amplifier.
Let us examine the circuit in Figure 5.63a to emphasize the similarity of the DC analysis for the PNP transistor to the NPN transistor case. Assuming a of 150 and a VEB voltage drop of 0.7 V, what is the expected voltage VC at the collector terminal? To answer this question, we can begin by calculating the emitter current based on the assumed VEB voltage drop, which gives iE = (VCC -1V + VEB) / RE = 0.588 mA. Solving equation 5.6 for α with = 150 results in α = 0.9934. Hence, IC = α·IE = 0.584 mA and VC = IC·RE = 1.17V. To check that the transistor is biased in active region, we can calculate VEC = (VCC - iE·RE) - VC = 3.53 V. Since VEB is larger than 0.3 V, the transistor is in active region.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
55
Now consider the circuit in Figure 5.63b, in which the base terminal of the additional NPN transistor is connected to node VC. Since there is no DC blocking capacitor between the transistors, the NPN transistor will affect the voltage at the collector of the PNP transistor. The calculation of current iC = 0.584 mA remains the same as in the previous example, but the voltage VC is now determined by the reduced current flowing through resistor RC. The equation that we can write for this voltage is VC = (iC – iB2)·RC. Assuming = 150 and VEB = 0.7 V for the NPN BJT, its base current can be expressed as iB2 = iC2/ = α·iE2/= α·(VC - 0.7V)/(·RE2). Solving these two equations simultaneously for their two unknown variables yields iB2 = 81.0 µA and VC = 1.01 V. The addition of the second stage has led to a VC reduction by 0.16 V. To avoid such loading effects, the multi-stage amplifier discussed in the previous section has the DC blocking capacitors between stages, which also simplify the DC analysis since each stage can be analyzed individually.
(a) (b)
Fig. 5.63. Example circuits for DC analysis.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
56
Appendix 5A: Harmonic distortion of an emitter-degenerated common-emitter amplifier. Let us consider the circuit in Figure A5.1 again. Although it is not entirely correct, let us assume that the voltage at the base vB is a linear combination of a DC bias voltage VB (generated from VCC, R1, R2) and the AC input signal vi, so that it can be expressed as vB = VB + vb.
Fig. A5.1. Common-emitter amplifier with emitter degeneration resistor.
Notice that vB is measured from the base terminal to ground. The collector current is generated by the base-emitter voltage (vBE = vB-vE = vB - iERE vB - iCRE) and it is given by
CQV
Riv
V
Riv
V
RIV
SV
RiIvV
SV
Riv
SC IeeeIeIeIi th
Ecb
th
Ecb
th
ECQB
th
EcCQbB
th
ECB
)(
, (A5.1)
where it is assumed that is high enough to make the substitution = 1 with negligible error. This equation can also be rewritten in the following form by taking the natural logarithm of both sides:
CQ
c
CQ
cCQ
CQ
C
th
Ecb
I
i
I
iI
I
i
V
Riv1lnlnln , (A5.2)
where iC is expressed as an addition of the DC component ICQ and the non-linear AC component ic. The function ln(x) stands for the natural logarithm of x. For small x, the term ln(1+x) can be expanded in a Taylor series as ln(1+x) = x - 0.51x2 + 0.34x3 - … Expansion of natural logarithm term in equation A5.2 with this series leads to:
..3
1
2
1..34.051.0
3232
CQ
c
CQ
c
CQ
c
CQ
c
CQ
c
CQ
c
th
Ecb
I
i
I
i
I
i
I
i
I
i
I
i
V
Riv (A5.3)
To analytically solve this equation is cumbersome and very often numerical solutions are used. For the case ic < ICQ, the high-order terms are small enough to be ignored, and equation simplifies to
32
3
1
2
11
CQ
c
CQ
c
CQ
c
th
CQE
th
b
I
i
I
i
I
i
V
IR
V
v. (A5.4)
This is a non-linear equation that is difficult to solve. Numerical methods or circuit simulations can of course provide the solutions. However, it is helpful to find approximate solutions that help us to properly use the degrees of freedom we may have during the design of a circuit. For this purpose, a convenient approximation approach is
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
57
introduced here, but the results should still be verified by simulating the design. If the collector current ic is assumed to have three dominant harmonic components, then it can be written as
3
c032
c02c0c ikikii . (A5.5)
where ic0 is the fundamental component of the collector current, and the currents ic0
2 and ic03 are the result of second
and third order non-linearities present in the amplifier. If we assume that the input signal vb consists of an ideal tone as shown in topmost spectrum of Figure A5.2, then the output current ic consists of several harmonic components as depicted in the same figure. Since the spectrum on the top is related to the ideal (perfectly linear) input tone, it only has a component at the fundamental frequency ωo, but not at other frequencies. The spectral components of the terms on the right side of equation A5.4 must cancel each other at all frequencies in order to preserve mathematical equivalence after the substitution of A5.5 for ic in A5.4, resulting in only one remaining component at ωo. This mathematical correspondence is visualized through the three spectra on the bottom of Figure A5.2, which represent the right side of equation A5.4. In this mathematical model, the input signal is assumed to be ideal, but perfectly linear signal sources do not exist in the real world. Nevertheless, the linearity of the signal source can be much better than the linearity of the circuit under characterization, which allows direct measurement of the circuit’s linearity characteristics in many cases. To consider a signal source “ideal”, its harmonic distortion components should be significantly lower than those of the circuit under investigation. This can be ascertained by measuring the output spectrum of the signal source prior to applying the test signal to a circuit. The harmonic distortion components of the circuit under test should be at least several decibels higher than those in the measurement of the signal source, otherwise you have to acquire a signal generator with better linearity in order to be able to measure the circuit’s distortion components.
th
b
V
v
CQ
c
th
CQE
I
i
V
IR1
2
CQ
c
I
i
2
1
3
CQ
c
I
i
3
1
Fig. A5.2. Graphical interpretation of equation A5.4. The AC input voltage signal vb consists of a single tone. Currents ic, ic
2, and ic3 consists of fundamental tones at = o and several additional harmonic components.
Substituting equation A5.5 into equation A5.4 yields
..ikiki
3
1
ikiki
2
1ikiki1
33c03
2c02c0
23c03
2c02c0
3c03
2c02c0
CQ
CQCQth
CQE
th
b
I
IIV
IR
V
v, (A5.6a)
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
58
which can be rearranged to
3
c02
23
2
c02
c0
i..
3
1kk1
i..
2
1k1
i1
CQCQCQ
th
CQE
CQCQ
th
CQE
CQth
CQE
th
b
III
V
IR
II
V
IR
IV
IR
V
v . (A5.6b)
Although it is very difficult to find the exact solution for this equation, some observations can be made that lead to very simple and relatively accurate results. First at all, the spectrum of the left hand side term is proportional to vb, and in a first approximation it consists of a single tone if an ideal sinusoidal is applied to the amplifier. Hence, the combination of all spectral components on the right hand side of equation A5.6 must consist of a single tone as well because the two sides (spectrums) have to be equal. According to equation A5.6, the right hand side coefficients of this expression must sum up to zero with the exception of the first one that defines the small-signal transconductance of the amplifier. Therefore, it follows that
CQth
CQE
th
b
Iv
IR
V
v c0i1 , (A5.7a)
02
1k1 2
CQ
th
CQE IV
IR , (A5.7b)
03
1kk1 2
23
CQCQ
th
CQE IIV
IR. (A5.7c)
From A5.7a, the relationship between the AC base voltage and the AC collector current at the fundamental signal frequency becomes
bEm
mb
th
CQE
th
CQ
vRg
gv
V
IRV
I
1
1
ic0 , (A5.8)
which fits with the small-signal analysis result for this common-emitter amplifier with emitter degeneration resistor RE. As expected, the effective small-signal transconductance with emitter degeneration is gm/(1 + gmRE). Taking the equivalent load resistance RC||RL and the input voltage divider zi/(zi + RS) into account, the overall small-signal gain is obtained as
Em
LCm
Si
ivv
Rg
RRg
Rz
zi
o
1
)||(. (A5.9)
This expression is similar to the one given in Table 5.1, which is written in terms of re = α/gm. From equations A5.7b and A5.7c, the non-linearity coefficients k2 and k3 can be expressed as
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
59
CQth
CQE IV
IR1
1
2
1k2
, (A5.10a)
22
2
23
1
21
6
1
1
1
3
k31k
CQth
CQE
th
CQE
CQth
CQE
CQ
IV
IR
V
IR
IV
IR
I. (A5.10b)
Therefore, the second and third harmonic distortions with a peak AC collector current amplitude of ic-pk can be approximated as
CQ
pkc
th
CQEpkc I
i
V
IRi
1
1
4
1
2
kHD2 2 , (A5.11)
2
223
1
21
24
1
4
kHD3
CQ
pkc
th
CQE
th
CQE
pkc I
i
V
IR
V
IR
i . (A5.12)
These equations can be used as a tool to define the ratio of the DC collector current to the expected collector current variations as function of the harmonic distortion components. Notice in this equation that the linearity is a strong function of RE, which determines the amplifier’s input impedance; the larger RE the better the linearity is, and the better the accuracy of the voltage gain. The drawback is that is the limitation in voltage gain. Equations A5.11 and A5.12 can be verified by considering the example II in section V.10. The amplifier shown in Fig. 5.51 is able to handle input signal up to 100mVpk while HD2=-40dB and HD3=-62dB according to Fig. 5.53. For ICQ=350A and RE=675Ω, eqn A5.11 predicts that HD2=-40dB if we limit the peak value of the collector current such that ic-pk0.4ICQ. For this case, Eqn A5.12 predicts HD3=-64dB, which fits well with the -62dB predicted by SPICE. Fig. A5.3a shows the collector current: ic-pk=135Apk. When the input signal is increased up to 200mVpk, ic-pk=260Apk as displayed in Fig. A5.3b. For this case, Equation A5.11 predicts HD2=-34.6 dB, which again fits well with SPICE results. HD3 is computed from A5.12 as -48dB, which again fits well with the Spice results shown in Fig. 5.54.
ELEN-325. Introduction to Electronic Circuits: A Design-Oriented Approach Jose Silva-Martinez and Marvin Onabajo
60
(a)
(b)
Fig. A5.3. Collector current measured for the case: a) vin-pk=100mV and b) vin-pk=200mV. The theoretical bias current ICQ=350A while SPICE results show ICQ=370A.